You are on page 1of 7

JOURNAL OF CHEMICAL PHYSICS VOLUME 114, NUMBER 17 1 MAY 2001

Adsorption state of dimethyl disulfide on Au„111…: Evidence for adsorption


as thiolate at the bridge site
Tomohiro Hayashia)
National Institute of Materials and Chemical Research, 1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan and
Institute of Materials Science, University of Tsukuba, Tsukuba, Ibaraki 305-8573, Japan
Yoshitada Morikawab)
Joint Research Center for Atom Technology (JRCAT), National Institute for Advanced Interdisciplinary
Research (NAIR), 1-1-4 Higashi, Tsukuba, Ibaraki 305-8562, Japan and School of Materials Science,
Japan Advanced Institute of Science and Technology (JAIST), 1-1 Asahidai, Tatsunokuchi,
Ishikawa 923-1292, Japan
Hisakazu Nozoyec)
National Institute of Materials and Chemical Research, 1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan and
Institute of Materials Science, University of Tsukuba, Tsukuba, Ibaraki 305-8573, Japan
共Received 9 January 2001; accepted 9 February 2001兲
We studied the adsorption state of dimethyl disulfide and methylthiolate on the Au共111兲 surface by
means of the density functional theory 共DFT兲 within a generalized gradient approximation and
experimental high-resolution electron energy loss spectroscopy 共HREELS兲 techniques. It turns out
that the methylthiolate adsorption is more stable than the dimethyl disulfide adsorption and that the
most stable adsorption site for the methylthiolate is the bridge site slightly off-centered towards the
fcc-hollow site with its S–C bond tilted from the surface normal by 53°. HREELS results are in
excellent agreement with the DFT results, providing very strong support to the depicted adsorption
scenario. © 2001 American Institute of Physics. 关DOI: 10.1063/1.1360245兴

I. INTRODUCTION two S atoms are adsorbed at different adsorption sites, ac-


cording to the results of the x-ray standing wave and the
Due to stability and high ordering, self-assembled mono-
grazing incidence x-ray diffraction techniques.11,12 Kluth
layers 共SAMs兲 have enormous potential for various applica-
et al. stated that the molecules are adsorbed as thiolate at
tions such as coating material, sensors, nonlinear optical ma-
room temperature forming nonequilibrium structures and
terials, catalysis, molecular devices, and so on.1–4 Among
that the dimerization is thermally activated at 375 K.13
them, alkanethiols (CH3 (CH2 ) n SH) and dialkyl disulfides
On the theoretical front, classical molecular dynamics
(CH3 (CH2 ) n S–S(CH2 ) n CH3 ) on Au共111兲 surfaces have
simulations assuming the thiolate adsorption14–16 cannot pre-
been the most widely studied SAM systems because they are
dict a specific structure for the experimentally observed
simple and easy to prepare even in air or in solution, as well
c(4⫻2) superlattice,17–19 while those assuming disulfide ad-
as in vacuum.5–7 However, the adsorption states of these
sorption were in good agreement with experimental results.20
molecules, especially, the interaction between the S head
Ab initio cluster calculations, on the other hand supported
groups and the Au substrates, are still a matter of intense
methylthiolate adsorption on Au clusters.21–23 Very recently,
debate.
Grönbeck et al. examined methylthiol, methylthiotate, and
From the results of high-resolution electron energy loss
DMDS adsorption on the Au共111兲 surface using density
spectroscopy 共HREELS兲, thermal desorption spectroscopy
functional theory with slab models. They concluded that the
共TDS兲, and x-ray photoelectron spectroscopy 共XPS兲 experi-
S–S bond of dimethyl disulfide is cleaved and methylthiolate
ments, Nuzzo et al. concluded that the S–S bond of dimethyl
is adsorbed at the fcc-hollow site.24
disulfide (CH3 SSCH3 , denoted by DMDS hereafter兲 is
The adsorption phases of alkyl thiolate monolayers are
cleaved at room temperature and the molecule is adsorbed as
governed by the van der Waals interaction among alkyl
thiolate with the S–C bond tilted from the direction normal
chains as well as by the bonding interaction between the S
to the surface.8 Ishida et al.9 and Noh et al.10 also supported
headgroup and the Au substrate. Although the van der Waals
the cleavage of the S–S bond based on their observations of
interaction among alkyl chains can be well described by em-
phase-separated domains in SAM films of asymmetric disul-
pirical force field in classical molecular dynamics simula-
fide on Au共111兲. Contrary to the above studies, Fenter et al.
tions, the interaction between the S headgroup and the Au
suggested that decanethiol is adsorbed as disulfide and that
substrate cannot. In this work, we studied the adsorption
state of DMDS and methylthiolate (CH3 S) on the Au共111兲
a兲
Present address: Institute für Angewandte Physikalische Chemie, Univer- surface using theoretical as well as experimental techniques
sität Heidelberg, Im Neuenheimer Feld 253, D-69120 Heidelberg, Ger-
to clarify the interaction between the S headgroup and the
many. Electronic mail: tomohiro.hayashi@urz.uni-heidelberg.de
b兲
Electronic mail: morikawa@jrcat.or.jp Au substrate. These molecules have minimum length of alkyl
c兲
Electronic mail: nozoye@nimc.go.jp chains and therefore, they are suitable to elucidate the nature

0021-9606/2001/114(17)/7615/7/$18.00 7615 © 2001 American Institute of Physics

Downloaded 02 Jun 2003 to 147.142.186.53. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
7616 J. Chem. Phys., Vol. 114, No. 17, 1 May 2001 Hayashi, Morikawa, and Nozoye

of the S–Au bonding. In order to analyze the adsorption TABLE I. The equilibrium lattice constant (a 0 ), the bulk modules B 0 , and
the cohesive energy (E c ) of the fcc Au. Previous calculations using full
states and the vibrational modes of adsorbates, we have car-
potential linearized augmented plane wave 共FLAPW兲 and pseudopotential
ried out first-principles calculations with repeated slab mod- 共PP兲 method and experimental data are compared.
els, which should be more suitable to mimic the ordered
molecular adsorption on metal surfaces than cluster models Property This work FLAPWa PP Expt.
used in previous studies. The HREELS measurements enable a 0 共Å兲 4.15 4.16 4.15 b
4.08e
us to observe low-frequency vibrational modes, which sup- B 0 共GPa兲 137 142 173.2e
ply information about the adsorption state of molecules, such E c 共kcal/mol兲 70.6 72.6b 87.96e
as the Au–S stretch mode. We find an excellent agreement of
␥ 100 共J/m2 ) 0.90 0.82c
the computed and the experimental HREEL spectra, thus
␥ 111 共J/m2 ) 0.72 0.76d 1.50f
providing very strong support to our results.
a d
Reference 39. Reference 41.
b e
II. COMPUTATIONAL AND EXPERIMENTAL DETAILS Reference 24. Reference 42.
c f
Reference 40. Reference 43.
All our calculations are carried out using a program
package STATE 共Simulation Tool for Atom Technology兲
which has been successfully applied for semiconductor as
The chamber is equipped with Auger electron spectroscopy
well as metal surfaces.25–27 Calculations are based on the
共AES兲, low-energy electron diffraction 共LEED兲 optics, quad-
density functional theory 共DFT兲,28,29 within a generalized
rupole mass spectrometer, scanning tunneling microscopy
gradient approximation 共GGA兲 using the Perdew–Burke–
共STM兲, and HREELS. The specimen is spark cut from a
Ernzerhof 共PBE兲 exchange-correlation energy functional.30
single crystal rod to face the 共111兲 surface and polished by a
All pseudopotentials are generated from scalar relativistic31
conventional method. The Au共111兲 surface is cleaned by
all electron atomic calculations to include relativistic effects
Ar⫹ sputtering and annealing 共900 K兲 procedures. DMDS is
which are important for heavy elements like Au. The spin–
introduced into the vacuum chamber through a variable leak
orbit interaction is omitted in this approximation but it is an
valve. The primary energy of electron beam of HREELS
good approximation for ground state properties of Au con-
used in this work is 4 eV and estimated sample current is
taining systems.32 The Vanderbilt’s ultrasoft pseudopotential
below several nA. In these conditions, we have checked that
scheme33 is used for H 1s, C 2p, and Au 5d and the norm
the electron beam does not damage the molecule on the sur-
conserving pseudopotential scheme34 is used for other states.
face. The spectra are taken at the specular scattering geom-
Valence electrons are expanded in a plane wave basis set.
etry. The resolution of the HREEL spectrum is 12–20 cm⫺1 .
The cut-off energies for the wave function and the augmen-
tation charge are 25 Ry and 225 Ry, respectively, and for
some systems we increased them to 30.25 Ry and 400 Ry,
III. RESULTS AND DISCUSSION
respectively, to check the accuracy of the calculations. For
wave function optimization, the block Davidson scheme is A. Bulk, clean surface, and isolated molecules
used.35 For the charge density mixing, the higher order To demonstrate the accuracy and the limitations of our
Anderson method proposed by Blügel36 is used and the gen- method, we describe some test results of bulk, clean surfaces,
eralized direct inversion in the iterative subspace 共GDIIS兲 and molecules in this subsection. Calculated bulk properties
scheme is used for the structure optimization. The Au共111兲 of fcc Au and surface energies of Au共100兲 ( ␥ 100) and
surface is simulated by a repeated slab model, in which four Au共111兲 ( ␥ 111) are summarized in Table I along with previ-
atomic layer slabs are separated by vacuum region of 14.4 Å. ous calculations24,39–41 and experimental42,43 results. Our re-
Adsorbates are introduced on one side of each slab and the sults agree well with the previous GGA results. However,
work function difference between the two sides of the slab is GGA underestimates the bulk cohesive energy by 20% and
compensated by introducing a dipole layer in the vacuum
region.37 The effect of this dipole correction is actually very
small. Therefore, we omitted the correction during the struc-
TABLE II. The equilibrium bond distances (r e ), vibration frequencies
tural optimization and it is included to calculate the dynamic ( ␻ e ), dissociation energies (D e ), and dipole moments ( ␮ ) of Au2 and AuH.
dipole moments of adsorbate vibrational modes. The adsor- Previous calculations using a scalar relativistic scheme by the elimination of
bates and the first layer substrate atoms are then fully re- the small components 共RESC兲 and experimental data are compared.
laxed. We used uniform k-point mesh for Brillouin zone
Molecule Property This work RESCa Expt.b
sampling and the Fermi level is smeared using first-order
Methfessel and Paxton scheme.38 With these computational Au2 r e 共Å兲 2.513 2.521 2.472
conditions, we have checked that the energy differences ␻ e (cm⫺1 ) 173 182 191
D e 共kcal/mol兲 53.7 59.3 53.3
among different adsorption structures are converged with a
numerical accuracy of ⬃2 kcal/mol per one CH3 S. The vi- AuH r e 共Å兲 1.540 1.533 1.524
brational mode analysis is carried out by diagonalizing the ␻ e (cm⫺1 ) 2273 2294 2305
Hessian matrices obtained by displacing the first layer gold D e 共kcal/mol兲 73.1 78.9 77.5
atoms and adsorbate atoms by 0.015–0.1 Å. ␮ 共D兲 1.050 1.004
All our experimental works are carried out in a ultra high a
Reference 32.
vacuum chamber. The base pressure is below 2.0⫻10⫺8 Pa. b
Reference 44.

Downloaded 02 Jun 2003 to 147.142.186.53. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
J. Chem. Phys., Vol. 114, No. 17, 1 May 2001 Adsorption of dymethyl disulfide on Au(111) 7617

TABLE III. The equilibrium length of the S–S 关 r e (SS) 兴 and the S–C TABLE IV. The adsorption energy (E ad), the height of the S atom from the
关 r e (SC) 兴 bonds, the angle between the S–S and the S–C bonds 关 ⬔共SSC兲兴, ideal bulk terminated Au共111兲 surface (h S), the angle of the S–C bond with
the dihedral angle of the CSSC skeleton 关 ␦ 共CSSC兲兴, and the S–S bond respect to the surface normal ( ␪ S–C) and the nearest neighbor Au–S distance
strength 关 D e 共S–S兲兴 of isolated DMDS compared with the experimental re- 关 r共Au–S兲兴 for methylthiolate and DMDS adsorption on the Au共111兲 sur-
sults. face.

Property This work Expt. Site E ad 共kcal/mol兲 h S 共Å兲 ␪ S–C(°) r共Au–S兲 共Å兲

r e 共SS兲 共Å兲 2.057 2.038a


bridge 12.4 2.05 52.7 2.50
r e 共SC兲 共Å兲 1.822 1.810a 共shifted to fcc site兲
⬔SSC 共°兲 102.1 102.81a bridge 11.4 2.09 53.4 2.50
␦ 共CSSC兲 80.7 84.7a 共shifted to hcp site兲
D e 共S–S兲 64.0 65b fcc-hollow 1.3 1.78 16.7 2.52
dimer model 3.9 2.65, 2.97 62.7, 46.4 2.69, 3.13
a
Reference 45.
b
Reference 46.

ylthiolate adsorbed Au共111兲 surface, we use the 冑3


the surface energy by 52% compared with experimental re-
⫻ 冑3R30° unit cell, as suggested by LEED observation48
sults. As is well known, the local density approximation
共LDA兲 underestimates the equilibrium lattice constants of and for DMDS adsorption, we use the 3⫻ 冑3 unit cell. The
solids and overestimates the bulk moduli and cohesive ener- surface Brillouin zone is sampled at 64 共24兲 uniform mesh of
gies, while GGA tends to correct them. GGA correction k-points for the 冑3⫻ 冑3R30° (3⫻ 冑3) unit cell.
works quite well for 3d transition metals while it overshoots Table IV summarizes the adsorption energies and the
the experimental results for 5d transition metals.39 As for structural parameters of various adsorption geometries. Here,
molecular properties, we summarized in Table II, the com- the adsorption energy for methylthiolate is defined by
puted equilibrium bond distances (r e ), the vibrational fre- E ad⫽⫺ 关 2⫻ 共 E 共 CH3S/Au共 111兲兲 ⫺E 共 Au共 111兲兲兲
quencies ( ␻ e ), and the dissociation energies (D e ) of Au2
and AuH molecules and in Table III, the molecular structures ⫺E 共 CH3SSCH3兲兴 , 共3.1兲
and the S–S bond energy for DMDS along with previous where, E(CH3S/Au(111)), E(Au(111)), and
theoretical32 and experimental44–46 data. In contrast to the E(CH3SSCH3) are total energies of CH3 S adsorbed
bulk cohesive energy and the surface energies, the bond Au共111兲, clean Au共111兲, and gas phase DMDS. To compare
strength of these molecules are quite well reproduced. These the adsorption energy of methylthiolate with that of DMDS,
results suggest that although the local adsorption geometry we included the factor 2 in the definition. The adsorption
of molecules on Au共111兲 may be reproduced reasonably energy for DMDS is similarly defined by
well, the binding energy between the S headgroup and the
Au substrate may have a seriously large error.47 We will E ad⫽⫺ 关共 E 共 CH3SSCH3/Au共 111兲兲 ⫺E 共 Au共 111兲兲兲
discuss the limitations of GGA for surface properties in some ⫺E 共 CH3SSCH3兲兴 . 共3.2兲
more detail later.
In the most 共second most兲 stable configuration, the S atom is
located at the bridge site slightly off centered towards the fcc
B. Adsorption geometry of methylthiolate 共hcp兲-hollow site and the S–C bond is tilted from the surface
and dimethyl disulfide at 1 ML
normal by 52.7°共53.4°兲 as shown in Fig. 1共a兲 共denoted as
We optimized adsorption structures of methylthiolate ‘‘bridge configuration’’ hereafter兲. In the third stable con-
starting from various initial configurations. For 1 ML meth- figuration, the S atom is located at the fcc hollow site and the

FIG. 1. 共Color兲 The top views and side views of 共a兲 the
bridge and 共b兲 the fcc-hollow configurations. 冑3
⫻ 冑3R30° unit cell used in our calculations is shown
by solid lines.

Downloaded 02 Jun 2003 to 147.142.186.53. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
7618 J. Chem. Phys., Vol. 114, No. 17, 1 May 2001 Hayashi, Morikawa, and Nozoye

S–C bond is tilted from the surface normal by 16.7° as TABLE V. The adsorption energy of methylthiolate at the bridge site
Bridge fcc
shown in Fig. 1共b兲 关denoted as ‘‘fcc hollow configuration’’ (E ad ) and the fcc hollow site (E ad ) as a function of the coverage (⌰),
the number of k-points (nk), and the number of layers (nL) in kcal/mol.
hereafter兴. We have also examined the hcp hollow configu- The adsorption energy is defined in the text.
ration with the S–C bond tilted from the surface normal by
16.7°. Then, the S–C bond is tilted further and the S atom ⌰共ML兲 nk nL bridge
E ad fcc
E ad bridge
E ad ⫺E ad
fcc

moved towards the bridge site during the geometry optimi- 1 16 4 15.2 7.0 8.2
zation. This clearly indicates that the hcp hollow configura- 1 36 4 12.7 1.3 11.4
tion is not a local minimum structure. Previous theoretical 1 64 4 12.4 1.3 11.1
calculations using cluster models suggested that the hollow 1 100 4 12.6 1.6 10.9
site adsorption was the most stable adsorption state.21,22 Our 1 64 3 9.7 ⫺0.5 10.1
1 64 5 12.1 1.7 10.3
calculations, however, indicate that the fcc-hollow site ad- 1 64 6 13.4 2.4 11.0
sorption is less stable than the bridge site adsorption by 5.5 1 64 7 11.5 0.8 10.7
kcal/mol. We believe that this discrepancy is probably due to 1 64 8 14.8 3.2 11.6
the artifact of the small cluster size used in calculations of 1/3 16 4 10.9 2.7 8.2
Refs. 21 and 22 and/or accuracy problems in the geometry 1/3 25 4 17.4 9.5 7.9
1/3 36 4 10.5 3.2 7.3
optimization. Recent DFT calculations using slab models 1/3 36 6 12.5 4.2 8.3
also concluded that the fcc-hollow configuration is the most 3/16 9 4 11.2 3.7 7.5
stable one.24 Furthermore, the adsorption energy of meth- 3/16 16 4 12.8 6.0 6.8
ylthiolate calculated by Grönbeck et al. 共38 kcal/mol兲 is 3/25 1 4 35.7 34.4 1.2
much larger than our value 共1.3 kcal/mol兲. We think that 3/25 4 4 25.9 24.3 1.6
3/25 9 4 21.4 16.6 4.8
these discrepancies presumably come from the different cov- 3/25 4 6 19.7 16.4 3.3
erage regimes treated in our and Grönbeck’s calculations and
we will discuss this point in the next subsection. A recent
study of methylthiolate adsorption on a Au38 nanocrystal
suggested the bridge site adsorption,23 being consistent with ber of slab layers are summarized. At 1.0 ML, the adsorption
our results. energies converge quite well using 64 k-points and 4 layer
As for DMDS adsorption, the geometry optimization is slab. As the coverage decreases, the adsorption energies get
carried out starting from the structural model proposed by more dependent on the number of k-points and number of
Fenter et al.,12 in which the initial positions of two sulfur slab layers, being difficult to obtain the converged results.
atoms are at the on-top site and at the fcc-hollow site and the Nevertheless, it is clear that the energy difference between
height of the sulfur atoms are 2.21 Å and 2.97 Å, respec- the bridge and the fcc-hollow configurations gets smaller as
tively. The DMDS molecule displaces far from the surface the coverage decreases and at ⌰⫽3/25 ML, it is only 1.2–
during the structure optimization, and the height of the sulfur 3.3 kcal/mol which is nearly the technical accuracy of our
atoms increased to 2.81 Å and 3.94 Å in the optimized struc- calculation. This may be the reason why Grönbeck et al.
ture 共denoted as ‘‘Fenter’s configuration’’ hereafter兲. Fur- concluded that the fcc-hollow site was the most stable ad-
thermore, Fenter’s configuration is less stable than the bridge sorption site at ⌰⫽3/25 ML. Another interesting aspect ex-
configuration by 11.2 kcal/mol per one disulfide molecule. tracted from Table V is that the adsorption energy tends to
Hence we conclude that Fenter’s configuration is unlikely for increase as the coverage is decreased.49 For example, the
DMDS adsorption. We also examined the adsorption geom- adsorption energy of methylthiolate in the fcc configuration
etry of DMDS starting from Grönbeck’s model which was is 34.4 kcal/mol at 3/25 ML using a single k-point and 4
given in Supporting Information of Ref. 24 关denoted as layer slab which is close to Grönbeck et al.’s result of 38
‘‘Grönbeck’s configuration’’ hereafter兴. Grönbeck’s configu- kcal/mol. However, the adsorption energy at low coverage
ration was slightly more stable than Fenter’s one by 2.7 kcal/ regimes depend significantly on the number of k-points and
mol, but it is still less stable than the CH3 S adsorbed in the number of layers, getting difficult to obtain the converged
bridge configuration by 8.5 kcal/mol. The adsorption energy value.
of DMDS reported by Grönbeck et al. 共16 kcal/mol兲 is again Table VI summarizes the coverage dependence of geom-
much larger than our calculated value of 3.9 kcal/mol. We etry for methylthiolate adsorbed at the bridge and the fcc
think this is also due to different coverage regimes treated in hollow sites. At both sites, the S atom tends to move to the
the two studies. From these results, we conclude that the surface and the Au–S distance tends to decrease as the cov-
methylthiolate adsorption in the bridge configuration is the erage decreases, which is in accordance with the strengthen-
most stable state and DMDS adsorption with its S–S bond ing of the Au–S binding energy.
intact is a metastable state at 1 ML.
D. HREEL spectra and vibrational mode analysis
C. Coverage dependence
Based on the results of DFT calculations, we have so far
For lower coverage regimes, we use 3⫻3, 4⫻4, and obtained that the methylthiolate adsorption at the bridge site
5⫻5 unit cells for 1/3, 3/16, and 3/25 ML methylthiolate is more stable than the DMDS adsorption with its S–S bond
adsorbed Au共111兲 surfaces, respectively. In Table V, the ad- intact. However, as indicated in Sec. III A, there is a rather
sorption energies of the bridge and the fcc-hollow configu- large error in the bulk cohesive energy and the surface ener-
rations as a function of the number of k-points and the num- gies of fcc Au. It was pointed out by Hammer and co-

Downloaded 02 Jun 2003 to 147.142.186.53. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
J. Chem. Phys., Vol. 114, No. 17, 1 May 2001 Adsorption of dymethyl disulfide on Au(111) 7619

TABLE VI. The height of the S atom from the ideal bulk terminated
Au共111兲 surface (h S), the angle of the S–C bond with respect to the surface
normal ( ␪ S–C), and the nearest neighbor Au–S distance (r共Au–S兲兲 of me-
thylthiolate adsorbed at the bridge site and the fcc hollow site as a function
of the coverage (⌰).

Site ⌰共ML兲 h S 共Å兲 ␪ S–C(°) r共Au–S兲 共Å兲

bridge 1 2.05 52.7 2.50


1/3 2.03 52.6 2.49
3/16 2.01 52.4 2.49
3/25 1.96 51.7 2.47

fcc-hollow 1 1.78 16.7 2.52


1/3 1.78 16.4 2.50
3/16 1.73 15.8 2.48
3/25 1.62 14.0 2.46

workers that the adsorption energies for atoms and molecules


on transition metal surfaces calculated by the PBE functional
overestimates the experimental values by ⬃0.5 eV.47 In ad-
dition, GGA energy sometimes gives qualitatively wrong
relative stability among different geometries.50 For example,
the PBE functional predicts that the most stable adsorption
site of CO on the Ni共100兲 surface is the hollow site, which is
never observed experimentally, while the least stable adsorp-
tion site is the on top site, which is experimentally observed.
Nevertheless, the CO stretching mode is quite well repro-
duced by the present GGA.51 Furthermore, recent studies
showed that the GGA can reproduce experimentally ob-
served infrared absorption spectra and HREEL spectra of
molecular adsorbates very well.25,27 Therefore, vibrational
studies would give very firm support for the structure models
proposed above and in this subsection, we will give results of
the HREELS and theoretical vibrational analysis.
Figures 2共a兲 and 2共b兲 show the HREEL spectra of 1.0
ML DMDS adsorbed Au共111兲 surface. In this experiment,
both dosing of DMDS and measurements were done at room
FIG. 2. 共Color兲 HREEL spectra at the coverage of 1.0 ML taken in the
temperature. The coverage is estimated from the intensity specular scattering geometry at room temperature along with calculated in-
ratio of Au and S in AES and the completion of one mono- tensity. Intensity calculated according to the bridge configuration is shown
layer is confirmed by thermal desorption spectroscopy in red and that obtained from the fcc-hollow configuration is shown in
共TDS兲. From the angular distribution of scattered electrons, green. 共a兲 The vibrational energy region around the Au–S stretch 关 ␯ 共Au–S兲兴
mode at 230 cm⫺1 . 共b兲 The higher vibrational energy region: the S–C
it turns out that the contribution of the dipole scattering stretch 关 ␯ 共S–C兲兴 at 680 cm⫺1 , the CH3 rock 关 ␳ (CH3 ) 兴 at 949 cm⫺1 , the
mechanism is dominating in specular scattering geometry for CH3 symmetric 关 ␦ s (CH3 ) 兴 at 1298 cm⫺1 , and the asymmetric deform
all the modes shown in the spectra. modes 关 ␦ a (CH3 ) 兴 at 1415 cm⫺1 .
We computed the vibrational energies of methylthiolate
adsorbed on the Au共111兲 surface and also the peak intensity
due to the dipole scattering mechanism. The absolute inten- be 2.8° from the half width at half maximum of the angular
sity of energy loss peaks from the dipole scattering mecha- distribution of the elastic peak. n s is the coverage of adsor-
nism is expressed as52 bates. F s ( ␪ˆ c ) is given by
I loss
I elastic
⫽ 冉 冊
ប 共 1⫺2 ␪ E 兲 1/2 d ␮
8a 0 ⑀ 0 E I cos ␪ I dQ
2
1
F 共 ␪ˆ 兲 n ,
␻s s c s
共3.3兲 F s 共 ␪ˆ c 兲 ⫽ 共 sin2 ␪ I ⫺2 cos2 ␪ I 兲
␪ˆ c 2
1⫹ ␪ˆ c 2
where, a 0 is the Bohr radius, ⑀ 0 is the vacuum permittivity,
␻ s is the normal frequency, d ␮ /dQ is the dynamic dipole ⫹ 共 1⫹cos2 ␪ I 兲 ln共 1⫹ ␪ˆ c 2 兲 . 共3.4兲
moment, which is calculated from the change in the work The calculated results are also shown in Figs. 2共a兲 and 2共b兲
function difference between the two sides of a slab. E I is the and compared to the experimental data. The intensity calcu-
primary energy of electron beam, and ␪ I is the incident angle lated from the bridge configuration is shown in red and that
of electron beam. ␪ E ⫽h ␻ s /2E I and ␪ˆ c ⫽ ␪ c / ␪ E , where ␪ c is obtained from the fcc-hollow configuration is shown in
the acceptance angle of the spectrometer and is estimated to green. The calculated normal coordinates are shown in Figs.

Downloaded 02 Jun 2003 to 147.142.186.53. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
7620 J. Chem. Phys., Vol. 114, No. 17, 1 May 2001 Hayashi, Morikawa, and Nozoye

relative intensity among the four modes, i.e., the S–C stretch
关 ␯ 共S–C兲兴, the CH3 rock 关 ␳ (CH3 ) 兴 , the CH3 symmetric de-
form 关 ␦ s (CH3 ) 兴 , and the CH3 asymmetric deform
关 ␦ a (CH3 ) 兴 modes is very well reproduced by the bridge con-
figuration, giving strong support to our results.
HREELS measurements at different exposures have
been also carried out and essentially the same spectra were
obtained, indicating that the adsorption state of methylthi-
olate does not depend on the coverage. As discussed in the
previous subsection, the adsorption energy depends on the
coverage rather significantly, but the bridge configuration is
the most stable configuration in all coverage regimes we
have examined, being fully consistent with the HREELS
measurements.
We have also calculated the vibrational modes for the
dimer configuration. The vibrational frequencies and the
peak intensity are completely different from those of meth-
ylthiolate shown in Fig. 2, excluding the existence of DMDS
at room temperature.

IV. CONCLUSION
The adsorption state of DMDS was studied with DFT-
GGA calculations and HREELS experiment. We conclude
that in the most stable adsorption state, the S–S bond of
DMDS is cleaved and DMDS is adsorbed as
methylthiolate.53 The sulfur atom of methylthiolate is located
at the off-centered bridge site towards the hollow site and the
S–C bond is tilted from the surface normal by 53°. This
structural model can account for the experimental HREEL
spectra very well.

ACKNOWLEDGMENTS
We thank Professor K. Terakura and Dr. M. Boero for
valuable discussions, suggestions and critical reading of the
manuscript. The numerical calculations were performed at
the computer centers of JRCAT and Tsukuba Advanced
Computing Center 共TACC兲. The present work is partly sup-
ported by New Energy and Industrial Technology Develop-
ment Organization 共NEDO兲 and the Grant-in-Aid for Scien-
tific Research on Priority Areas ‘‘Molecular Physical
Chemistry’’ from the Ministry of Education, Science, Sports,
and Culture.
FIG. 3. 共Color兲 Vibrational normal coordinates of 共a兲 the ␯ 共Au–S兲 ⫹ hin-
dered rotation [R兴, 共b兲 the ␯ 共S–C兲, 共c兲 the ␳ (CH3 ), 共d兲 the ␦ s (CH3 ), and 共e兲
the ␦ a (CH3 ) modes for the bridge site configuration and 共f兲 and 共g兲 the R ⫹ 1
J. D. Swalen, D. J. Allara, J. D. Andrade, E. A. Chandross, S. Garoff, J.
␯ 共Au–S兲 modes for the fcc-hollow configuration. The vibrational normal Israelachvili, T. J. McCarthy, R. Murray, R. F. Pease, J. F. Rabolt, K. J.
coordinates are scaled by 20 times in the figures. Wynne, and H. Yu, Langmuir 3, 932 共1987兲.
2
L. H. Dubois and R. G. Nuzzo, Annu. Rev. Phys. Chem. 43, 437 共1992兲.
3
A. Ulman, Chem. Rev. 96, 1533 共1996兲.
4
J. Chen, M. A. Reed, A. M. Rawlett, and J. M. Tour, Science 286, 1550
3共a兲–3共g兲, in which the vibrational normal coordinates are 共1999兲.
scaled by 20 times. The most prominent feature of the spec- 5
H. Kondoh, C. Kodama, H. Sumida, and H. Nozoye, J. Chem. Phys. 111,
trum is the sharp peak assigned to the Au–S stretch 1175 共1999兲.
6
关 ␯ 共Au–S兲兴 mode appearing at 230 cm⫺1 . The theoretical T. Hayashi, A. Fricke, K. Katsura, C. Kodama, and H. Nozoye, Surf. Sci.
427Õ428, 393 共1999兲.
calculation assuming the bridge model can reproduce the 7
G. E. Poirier, Langmuir 15, 1167 共1999兲.
single strong peak intensity as well as the vibrational energy 8
R. G. Nuzzo, B. R. Zegarski, and L. H. Dubois, J. Am. Chem. Soc. 109,
of this mode while the fcc-hollow model fails. The splitting 733 共1987兲.
9
T. Ishida, S. Yamamoto, W. Mizutami, M. Motomatsu, H. Tokumoto, H.
of the ␯ 共Au–S兲 mode for the fcc-hollow model is due to the Hokari, H. Azehara, and M. Fujihira, Langmuir 13, 3261 共1997兲.
coupling between the ␯ 共Au–S兲 and the hindered rotation 关 R 兴 10
J. Noh and M. Hara, Langmuir 16, 2045 共2000兲.
modes as shown in Figs. 3共f兲 and 3共g兲. For other modes, the 11
P. Fenter, A. Eberhardt, and P. Eisenberger, Science 266, 1216 共1994兲.

Downloaded 02 Jun 2003 to 147.142.186.53. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
J. Chem. Phys., Vol. 114, No. 17, 1 May 2001 Adsorption of dymethyl disulfide on Au(111) 7621

12
P. Fenter, F. Schreiber, L. Berman, G. Scoles, P. Eisenberger, and M. J. 33
D. Vanderbilt, Phys. Rev. B 41, 7892 共1990兲.
Bedzyk, Surf. Sci. 412Õ413, 213 共1998兲. 34
N. Troullier and J. L. Martins, Phys. Rev. B 43, 1993 共1991兲.
13
G. J. Kluth, C. Carraro, and R. Maoudian, Phys. Rev. B 59, R10449 35
D. M. Wood and A. Zunger, J. Phys. A 18, 1343 共1985兲.
共1999兲. 36
S. Blügel, Ph.D. thesis, Tech. Univ. Aachen, 1988.
14
W. Mar and M. L. Klein, Langmuir 10, 188 共1994兲. 37
J. Neugebauer and M. Scheffler, Phys. Rev. B 46, 16067 共1992兲.
15
A. J. Pertsin and M. Grunze, Langmuir 10, 3668 共1994兲. 38
M. Methfessel and A. T. Paxton, Phys. Rev. B 40, 3616 共1989兲.
16
R. Bhatia and B. J. Garrison, Langmuir 13, 765 共1997兲. 39
A. Khein, D. J. Singh, and C. J. Umrigar, Phys. Rev. B 51, 4105 共1995兲.
17
N. Camillone III, C. E. D. Chidsey, G. Liu, and G. Scoles, J. Chem. Phys. 40
B. D. Yu and M. Scheffler, Phys. Rev. B 56, R15569 共1997兲.
98, 3503 共1993兲. 41
M. C. Vargas, P. Giannozzi, A. Selloni, and G. Scoles 共private communi-
18
P. Fenter, P. Eisenberger, and K. S. Liang, Phys. Rev. Lett. 70, 2447 cation兲.
共1993兲. 42
C. Kittel, Introduction to Solid State Physics, 5th ed. 共Wiley, New York,
19
G. E. Poirier, Chem. Rev. 97, 1117 共1997兲. 1976兲.
20
J. J. Gerdy and W. A. Goddard, III, J. Am. Chem. Soc. 118, 3223 共1996兲. 43
W. R. Tyson and W. A. Miller, Surf. Sci. 62, 267 共1977兲.
21 44
H. Sellers, A. Ulman, Y. Shnidman, and J. E. Eilers, J. Am. Chem. Soc. K. P. Huber and G. Herzberg, Constants of Diatomic Molecules, Vol. IV,
115, 9389 共1993兲. in Molecular Spectra and Molecular Structure 共Van Nostrand Reinhold,
22
K. M. Beardmore, J. D. Kress, N. Grønbech-Jensen, and A. R. Bishop, New York, 1979兲.
Chem. Phys. Lett. 286, 40 共1998兲. 45
D. Sutter, H. Dreizler, and H. D. Rudolph, Z. Naturforsch. A 20A, 1676
23
H. Häkkinen, R. N. Barnett, and U. Landman, Phys. Rev. Lett. 82, 3264 共1965兲.
共1999兲. 46
J. M. Nicovich, K. D. Kreutter, C. A. van Dijk, and P. H. Wine, J. Phys.
24
H. Grönbeck, A. Curioni, and W. Andreoni, J. Am. Chem. Soc. 122, 3839 Chem. 96, 2518 共1992兲.
共2000兲. 47
B. Hammer, L. B. Hansen, and J. K. Nørskov, Phys. Rev. B 59, 7413
25
Y. Morikawa, K. Iwata, J. Nakamura, T. Fujitani, and K. Terakura, Chem. 共1999兲.
Phys. Lett. 304, 91 共1999兲. 48
L. H. Dubois, B. R. Zegarski, and R. G. Nuzzo, J. Chem. Phys. 98, 678
26
Y. Morikawa, K. Iwata, and K. Terakura, Appl. Surf. Sci. 169Õ170, 11 共1993兲.
共2001兲. 49
This property is independently obtained by authors of Ref. 41 共private
27
Y. Morikawa, Phys. Rev. B 63, 033405 共2001兲. communication兲.
28
P. Hohenberg and W. Kohn, Phys. Rev. 136, B864 共1964兲. 50
L. Mitas, J. C. Grossman, I. Stich, and J. Tobik, Phys. Rev. Lett. 84, 1479
29
W. Kohn and L. J. Sham, Phys. Rev. 140, A1133 共1965兲. 共2000兲.
30
J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 51
Y. Morikawa 共unpublished兲.
共1996兲. 52
H. Ibach and D. L. Mills, Electron Energy Loss Spectroscopy and Surface
31
D. D. Koelling and B. N. Harmon, J. Phys. C 10, 3107 共1977兲. Vibrations 共Academic, New York, 1982兲, p. 100.
32 53
T. Suzumura, T. Nakajima, and K. Hirao, Int. J. Quantum Chem. 75, 757 The cleavage of the S–S bond is also directly observed by HREELS
共1999兲. experiment: T. Hayashi and H. Nozoye 共to be published兲.

Downloaded 02 Jun 2003 to 147.142.186.53. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

You might also like