Spectral correlation in the adsorption of aliphatic mercaptans on silver and gold surfaces: Raman spectroscopic and computational study

Article abstract of DOI:10.1016/S0022-2860(98)00895-3

Spectral shifts of the v(CS) vibrations of several aliphatic mercaptans adsorbed on the Ag and Au electrode surfaces were measured with the surface- enhanced Raman scattering. Influence of the chemisorption mechanism on the spectral shift was investigated

Full text of DOI:10.1016/S0022-2860(98)00895-3

MOLSTR 10798
Journal of Molecular Structure 479 (1999) 83–92
Spectral correlation in the adsorption of aliphatic mercaptans
on silver and gold surfaces: Raman spectroscopic and
computational study
*
Seung Il Cho, Eun Sun Park, Kwan Kim , Myung Soo Kim
Department of Chemistry and Center for Molecular Catalysis, Seoul National University, Seoul 151-742, South Korea
Received 20 July 1998; accepted 12 November 1998
Abstract
Spectral shifts of the n(CS) vibrations of several aliphatic mercaptans adsorbed on the Ag and Au electrode surfaces were
measured with the surface-enhanced Raman scattering. Influence of the chemisorption mechanism on the spectral shift was
investigated through ab initio quantum mechanical calculation for CH3SH, CH3SϪ, CH3SNa, and CH3SMn (M ˆ Ag, Au and n ˆ
1–3). It was found that even though the 2e molecular orbital of methanethiolate makes the major contribution to its chemisorp-
tion, the 3a1 orbital is mainly responsible for the n(CS) spectral shift. Adsorption of thiolates on the roughened Ag electrode
surface seems to occur at the multiple sites such as the three-fold hollow sites while the number of the surface Au atoms
interacting effectively with thiolates seems to be less than that of the Ag atoms. ᭧ 1999 Elsevier Science B.V. All rights
reserved.
Keywords: Raman; Surface-enhanced Raman scattering; Aliphatic mercaptans; Ag; Au
1. Introduction
alkanethiolate was investigated by electron diffraction
[6] and scanning tunneling microscopy [7,8]. Forma-
p p
Adsorption of organic mercaptans on Ag and Au
surfaces is the focus of tremendous research interest in
relation to the preparation of well-organized organic
thin films with practical applications [1–5]. Even
though various spectroscopic tools are available to
investigate such surface-adsorbate systems, there is
none which is capable of providing all the answers
relevant to the structure and bonding of the composite
systems. This necessitates the use of several techni-
ques in combination to elucidate the nature of the
systems described before. For example, the structure
on Ag or Au surface formed by self-assembling of
tion of a 3 × 3R30Њ structure was widely reported.
Information on the orientation and order of poly-
methylene chains of the adsorbed alkanethiolates is
available from infrared spectroscopy [9–11]. The tilt
angles of ϳ 30Њ and ϳ 12Њ for monolayers on Au
[9,11] and Ag [10,11], respectively, were reported,
indicating a difference in the sulfur head group at
the two metals. However, more direct information
on the head group interaction is difficult to obtain
from infrared spectroscopy. This is owing to the
weak intensities of the C–S stretching (n(CS)) and
metal–sulfur stretching (n(MS)) bands and other
technical difficulties.
Surface-enhanced Raman scattering (SERS) which
is known to be particularly sensitive to the vibrations
* Corresponding author. Tel.: ϩ 82-2-880-6651; Fax: ϩ 82-2-
889-1568; e-mail: kwankim@plaza.snu.ac.kr
0022-2860/99/$ - see front matter ᭧ 1999 Elsevier Science B.V. All rights reserved.
PII: S0022-2860(98)00895-3

84
S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
of atoms in the immediate vicinity of Ag and Au
surfaces has also been applied to the investigation of
the adsorption of organic mercaptans [12–16]. It was
clearly shown by this technique that mercaptans
chemisorb on the metal surfaces dissociatively after
losing their thiol protons. Substantial red-shifts,
namely the decrease in the vibrational frequencies,
of the n(CS) modes were taken as an evidence for
the participation of sulfur in the adsorption. Various
theoretical efforts were made to understand the cause
for such red-shifts and hence their adsorption mechan-
isms [17–19]. A consistent correlation between the
spectral shift and the adsorption mechanism has not
emerged yet, however. In this work, we present our
recent results from the combined experimental and
computational investigations on the adsorption of
organic mercaptans on Ag and Au surfaces. A parti-
cular emphasis was placed on the consistent explana-
tion of the spectral shift in terms of the adsorption
mechanism.
saturated calomel reference electrode (SCE). A
working electrode was prepared by polishing with
sandpaper and with 0.05 mm alumina in distilled
water and then oxidizing with equal volume mixture
of 30% H2O2 and concentrated ammonia solution. The
electrode was roughened in 0.1 M KCl solution by
oxidation–reduction cycles. The electrochemical
condition recommended by Gao et al. [23] was
adopted. The electrode was immersed in 10Ϫ5 M thio-
late solutions for 30 s. Then, the electrode was
brought back into the electrochemical system. The
potential of the working electrode was maintained at
Ϫ 0.2 V vs. SCE during SER spectral measurements
even though the SER spectra were invariant with
respect to the applied voltage. The reason for using
separate solutions for adsorption and SER spectral
recording is to avoid contamination of the electroche-
mical system with thiolates.
A sample was irradiated with a He/Ne laser
(Research Electro-Optics Model LHRP-2001). The
laser power was ϳ 10 mW at the sample position.
Raman scattering was collected at 90Њ using a f/1.4
camera lens with 50 mm focal length and focused
onto the entrance slit of a Spex 1877E Triplemate
spectrometer equipped with a charge-coupled device.
The slit width was set at the spectral resolution of ϳ
2. Experimental
Samples investigated in this work are methanethiol,
ethanethiol, propanethiol, buthanethiol, 2-methyl-2-
propanethiol, and cyclohexanethiol, and their anions
in the aqueous solutions. Ordinary Raman spectra of
sodium, silver, and gold thiolate salts were also
recorded. The sodium salts of methanethiolate, etha-
nethiolate, propanethiolate, buthanethiolate, and 2-
methyl-2-propanethiolate were purchased from
Fluka and washed thoroughly with ether. Sodium
cyclohexanethiolate was prepared by mixing cyclo-
hexanethiol with NaOH in methanol. Silver thiolate
salts (RSAg) were prepared by mixing an excess of
sodium thiolate or thiol in methanol with AgNO3 [20].
The resulting precipitates were washed thoroughly
with methanol. Gold thiolate salts (RSAu) were
synthesized by the reduction of KAuCl4 with thiol
or thiolate in methanol [21]. The resulting precipitates
were washed thoroughly with methanol and ether to
remove disulfides. The metal-to-thiolate ratios were
1 : 1 in all the salts as determined by the Fajans
method [22] and the elemental analysis.
10 cmϪ1
.
A
commercial data system (Spex
DM3000r), which was modified to improve cosmic
ray rejection, was used for data acquisition and
processing.
3. Computational
Ab initio molecular orbital calculation was done for
CH3SH, CH3SϪ, CH3SNa, CH3SAg, CH3SAg2,
CH3SAg3, CH3SAu, CH3SAu2, and CH3SAu3 with
the GAMESS package [24]. The 6-31G(3p, 3d)
basis sets were used except for the gold and silver
atoms. In order to reduce the amount of computation
and to include some relativistic effects, the silver and
gold atoms were treated with the effective core poten-
tials of Stevens et al. [25] and the double or triple zeta
basis sets recommended by them. The geometries of
all the systems except for those containing two or
three metal atoms were optimized at the Hartree–
Fock (HF) level. For the latter, the metal–metal
The electrochemical system used to record SER
spectra consists of three electrodes, a silver or gold
working electrode, a platinum counter electrode, and a
˚
distance was fixed at 2.89 and 2.88 A for silver and

S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
85
The main interaction in the formation of the silver
salts was identified as between the 2e orbitals of
CH3SϪ and the LUMO of Agϩ or Ag2ϩ. The calculated
frequencies of the n(CS) mode followed the order
CH3SH Ͼ CH3SAg Ͼ CH3SAg2 Ͼ CH3SϪ, in contrast
with the experimental order CH3SH Ͼ CH3SϪ
Ͼ
SERS. However, it was found through calculation
that the n(CS) frequency of CH3SϪ–H2O was higher
than that of free CH3SϪ. Hence, the experimental
order CH3SϪ Ͼ SERS was attributed to the hydration
of CH3SϪ in aqueous solution. Another interesting
theoretical investigation on the adsorption of CH3SϪ
on silver was carried out by Lee et al. [18] using the
atom superposition and electron delocalization mole-
cular orbital (ASED-MO) calculation by modeling the
Ag(111) and Ag(100) surfaces as clusters of 41–55
atoms. In particular, the reduced overlap population
(ROP) of the chemisorption bond and the contribution
from each MO of CH3SϪ were calculated. It was
found that the chemisorption was mainly owing to
the 2e and 3a1 MOs. Noting that the contribution
from the 2e MO was larger by a factor of 3 than
that from 3a1, it was suggested that electron donation
from the 2e orbitals in the chemisorption would result
in the blue-shift of the n(CS) mode of the adsorbed
CH3SϪ compared to the free CH3SϪ. Then, our
previous calculation on the CH3SϪ–H2O system [17]
was invoked to explain the fact that the experimental
order of the n(CS) frequencies was opposite to the
calculated ones. An important point we would like
to emphasize here is that the 2e orbitals are essentially
nonbonding while 3a1 is bonding. Namely, even
though the 2e orbitals make a large contribution to
the chemisorption, it is possible that the C–S bond
strength is more affected by the electron donation
from 3a1. Also noteworthy in the work of Lee et al.
[18] is that ROPs on the hollow sites are larger than
those on the on-top sites both on the Ag(111) and
Ag(100) surfaces indicating stronger chemisorption
on the former.
Fig. 1. A schematic energy diagram of the molecular orbitals of
CH3SϪ.
gold, respectively, as taken from the bulk data [26],
and other geometrical parameters were fully opti-
mized. Harmonic vibrational frequencies were
calculated at the optimized geometries.
4. Results and discussion
Before getting into the details of the present results,
let us first review briefly the previous experimental
and computational investigations [17–19] on the
adsorption of aliphatic thiols, methanethiol in parti-
cular, on the Ag surface. For this purpose, a schematic
energy diagram of the molecular orbitals of CH3SϪ is
shown in Fig. 1. Herein, 1a1 is a bonding orbital
formed between the s orbitals of carbon and sulfur.
The 2a1 orbital is confined to the methyl group,
possessing an antibonding character with respect to
the C–S bond. The 1e’s are degenerate bonding orbi-
tals between the px and py orbitals of carbon and the 1s
hydrogen atoms. The 3a1 orbital is a bonding orbital
formed between the spz hybridized orbitals of carbon
and sulfur atoms. The 2e orbitals, which are the
highest occupied molecular orbitals (HOMO), are
nonbonding orbitals virtually of px and py of sulfur
with slight antibonding mixing from carbon px and
py. Finally, 4a1, which is the lowest unoccupied mole-
cular orbital (LUMO), is nearly a nonbonding orbital
composed of pz of carbon and s, pz, and dz2 of sulfur.
In our previous investigation on the adsorption of
methanethiol on a silver surface, ab initio calculations
were carried out for CH3SϪ, CH3SAg, CH3SAg2 and
other related molecules at the HF/6-31G* level [17].
The highest level calculation performed so far was
reported by Sellers et al. [19] on the adsorption of
CH3SϪ on Ag(100), Ag(111), Au(100), and Au(111)
surfaces. Ab initio RECP HF ϩ correlation with
second-order perturbation theory (MBPT2) was
adopted. The C–S axis of methanethiolate adsorbed
on the hollow sites of all the surfaces was found to be
perpendicular to the surfaces while it was tilted by

86
S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
Fig. 2. Ordinary Raman spectra of (a) aqueous CH3SϪ and (b)
sodium, (c) silver, and (d) gold salts of methanethiolate. The
metal salts were rorated at 6000 rpm to minimize laser damage.
The integration times were 2, 10, and 60 s for sodium, silver, and
gold methanethiolate salts, respectively. The fluorescence back-
ground was subtracted.
Fig. 3. SER spectra of methanethiolate on (a) Ag and (b) Au
electrode surfaces. The integration times were 100 and 1000 s,
respectively.
CH3SϪ may have caused such a misinterpretation.
On these grounds, we prepared and measured the
n(CS) frequencies of the sodium salts of aliphatic
thiolates. The reason for studying these compounds
was that the n(CS)’s of these compounds would not
be much affected by the presence of metal owing to
the ionic nature of the Na–S bond and that these
frequencies would provide better reference points to
compare with than those for RSϪ in aqueous solution
which suffers from complication owing to hydration.
Measurements were extended to several thiolate
compounds to get a comprehensive view. Also,
measurements were made for gold thiolate salts and
for thiolates adsorbed on gold surfaces for
comparison.
ϳ70Њ from the surface normal when the adsorption
occurred on the on-top sites. Adsorption on the hollow
sites was a little bit more stable than on the on-top
sites for all the surfaces investigated. For the same
type of surfaces, the force constants for the n(CS)
and n(MS) modes were larger on the Au surfaces
than on Ag. However, the calculated results reported
are not sufficient for the investigation of the spectral
shifts upon surface adsorption.
The literature review shows that the quantum
mechanical calculations present a consistent view on
the chemisorption scheme regardless of the level of
the calculations. This suggests that the misinterpreta-
tion of the calculated results, not the results them-
selves, may be responsible for the unsatisfactory
explanation of the spectral shifts. Lack of experi-
mental data such as the n(CS) frequency of free
Ordinary Raman spectra of the sodium, silver, and
gold salts of methanethiolate are shown in Fig. 2
together with the one for CH3SϪ in basic solution.
The surface-enhanced Raman (SER) spectra of
CH3SϪ adsorbed on Ag and Au electrode surfaces

S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
87
Table 1
C–S stretching frequencies of aliphatic thiols
Compound
Ordinary Raman
RSH
SERS
Ag
RSϪ
RSNa
RSAg
RSAu
Au
Methanethiol
Ethanethiol
705
658
651
733
655
731
735
590
695
656
650
728
653
734
732
589
699
663
653
732
651
724
733
585
695
645
—
716
—
718
729
577
687
646
637
720
643
718
724
575
679
630
624
697
629
697
712
548
682
643
636
707
635
702
721
560
Propanethiol (G)a
Propanethiol (T)a
Butanethiol (G)a
Butanethiol (T)a
Cyclohexanethiol (Eq.)b
2-Methyl-2-propanethiol
a T and G denote the rotational isomers with trans and gauche conformations around the C(1)–C(2) bond, respectively.
b Eq. denotes the equatorial conformation.
are shown in Fig. 3. The C–S stretching frequencies in
these spectra and those in the corresponding spectra of
other aliphatic thiolates are listed in Table 1. It is to be
noted that the order of the C–S stretching frequency,
RSH Ͼ RSϪ Ͼ RSM Ͼ SERS (where M is Ag or Au)
is observed for almost all the thiolates investigated.
The n(CS) frequencies of the sodium salts are invari-
ably higher than the corresponding frequencies of the
silver or gold salts. These are higher than those of RSϪ
in the aqueous solution in some cases and lower in
others. This may be owing to the difference in the
extent of hydration of RSϪ in aqueous solution as
was suggested previously [17]. For example, the
n(CS) frequency of CH3SϪ in the aqueous solution
is lower than that of CH3SNa. Also noteworthy is
the fact that the red-shifts for the Ag salts are compar-
able to or less than those for the corresponding Au
salts. In contrast, the red-shifts on the Ag surfaces are
invariably larger than those on the Au surfaces. The
order of the n(CS) frequencies can be summarized as
RSNa Ͼ RSAg Ն RSAu Ͼ RS-Au surface Ͼ RS-Ag
surface. This result cannot be explained based on the
previous chemisorption scheme of thiolates described
before [17,18], necessitating further computation and
interpretation of the computed results.
Structural data and n(CS) and n(MS) frequencies
of various CH3SMn (M ˆ Na, Ag, Au and n ˆ 1, 2, 3)
species obtained from ab initio molecular orbital
calculations at the HF/6-31G(3p, 3d) level are listed
in Table 2. The calculated harmonic frequencies
Table 2
Structure and frequencies of the CH3S systems calculated at the HF level with 6-31G(3p, 3d) basis seta
AgSCH3
Ag2SCH3
Ag3SCH3
AuSCH3
Au2SCH3
Au3SCH3
NaSCH3
rSXb
rCS
rCHtr
rCHg
ЄCSX
ЄHtrCS
ЄHgCS
ЄHtrCSX
ЄHgCSX
C–S freq.
M–S freq.
2.4378
1.8283
1.0836
1.0828
106.16
107.51
111.70
180.00
61.37
2.2228
1.8329
1.0824
1.0818
113.08
108.25
111.31
180.00
60.81
2.8709
1.8316
1.0828
1.0828
144.47
110.00
110.00
180.00
60.00
2.3273
1.8247
1.0840
1.0812
104.94
106.24
111.61
180.00
61.70
2.1184
1.8324
1.0822
1.0801
111.96
107.56
110.95
180.00
60.91
2.8828
1.8299
1.0831
1.0831
144.77
109.95
109.95
180.00
60.00
2.5277
1.8298
1.0837
1.0844
111.24
108.79
111.64
180.00
61.00
750
298
746
225
735
160
748
342
740.5
241
732.5
157
752
310
a
Ϫ1
˚
Bond lengths (r) are in A, angles (Є) are in degrees, and frequencies are in cm
.
b X ˆ H for CH3SH; X ˆ Na, Ag, and Au for NaSCH3, AgSCH3, and AuSCH3, respectively; X is the center of the two metal atoms for
Ag2SCH3 and Au2SCH3; X is one of the three Ag or Au atoms for Ag3SCH3 or Au3SCH3.

88
S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
Fig. 4. Two-dimensional HF MO contour diagrams for CH3SAg and CH3SAu. In each plot, the contour spacing is 0.05 a.u. Solid lines denote
contours of positive amplitude and dashed lines denote those of negative amplitude or nodal planes. The plotting plane contains four atoms, i.e.,
M (Ag or Au), S, C, and Htrans, which are connected by solid lines. The C–Hgauche bond is also shown by solid lines. (a) 15A0 and (b) 12A0 MOs
for the CH3SAg system. (c) 15A0 and (d) 12A0 MOs for CH3SAu system.

S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
89
would be ϳ 10% higher than the real values, as is
usual for the calculation at this level. It is to be borne
in mind that the calculations are for CH3SMn mole-
cular species while multiple interaction is possible
between CH3SϪ and metal atoms in salt. Unfortu-
nately, detailed structural information on Ag and Au
salts of CH3SϪ from X-ray diffraction study is not
available. In the case of silver hexanethiolate, an
X-ray diffraction study shows that both the sulfur and
silver are either two-fold or three-fold coordinated
[27]. A closer inspection of the Ag–S distances
shows, however, that the interaction of some pairs
of Ag and S atoms is more important than others.
No crystal structure information is available for gold
thiolate salts. Considering the stoichiometry of salts,
we would assume that sulfur atoms interact with one
or two Ag or Au atoms in the salts. Namely, the C–S
bonds in the salts may be closer to those in CH3SM or
CH3SM2 species considered in the calculation rather
than those in CH3SM3. It is informative at this point to
compare the n(MS) modes which appear at 250 and
297 cmϪ1, respectively, in the spectra of silver and
gold salts of methanethiol (Fig. 2). The experimental
frequency difference of 47 cmϪ1, the n(AuS) being
higher than the n(AgS), compares well with the corre-
sponding difference between CH3SAu and CH3SAg in
the calculation, suggesting that the CH3SAu and
CH3SAg species may approximate the respective
salts rather well.
It is interesting to note the order in the experimental
n(CS) frequency of Na salt Ͼ Ag salt Ͼ Au salt. A
similar trend, CH3SNa Ͼ CH3SAg Ͼ CH3SAu, is seen
in the calculated results also. According to the
previous chemisorption model described earlier
[17,18], the electron donation from the 2e orbitals of
CH3SϪ to Ag or Au would result in the blue-shift of
the n(CS) mode owing to the slight anti-bonding char-
acter of these orbitals with respect to the C–S bond.
This is contrary to the experimental and computa-
tional findings. A feasible explanation for the red-
shifts is to assume that the electron donation from
the 3a1 orbital of CH3SϪ which is bonding with
respect to the C–S bond is mainly responsible for
the n(CS) frequency shift, even though the 2e orbitals
contribute more to the chemisorption bond. As an
attempt to obtain further support for this model, we
compared the contours of 12A0 and 15A0 molecular
orbitals of CH3SAg and CH3SAu species which are
formed by the interaction between 3a1 of CH3SϪ and d
orbitals of the metals. The contour diagrams in Fig. 4
show that these orbitals are more delocalized in
CH3SAu than in CH3SAg. Namely, its contribution
to chemisorption is larger in CH3SAu than in
CH3SAg, in agreement with the lower calculated
n(CS) frequency in CH3SAu than in CH3SAg. It is
worth mentioning that the 3a1 orbital in the sodium
salt is virtually the same as in the free anion, hardly
affected by the presence of Naϩ.
As was mentioned earlier, the red-shifts of the
n(CS) mode on the electrode surfaces are larger
than in the salts. In addition, the red-shift on the silver
surface is larger than that on the gold surface.
According to Sellers et al. [19], the bonding of the
sulfur atom at the on-top site of various surfaces is
concentrated upon a single metal atom. Then, the
metal salts considered before would approximate the
bonding at these sites rather well. The fact that
the red-shifts are larger on the surfaces than those in
salts suggests that the adsorption on the surfaces
occurs at the multiple sites such as two-fold or three-
fold sites. In this respect, some knowledge is needed
on the structures of the metal surfaces used. Considering
the microscopic roughness needed for a successful
SERS investigation, the presence of a mixture of
surface planes would not be surprising, however.
According to the investigations with the scanning
tunnelling microscopy, the surface structures of Ag
and Au metals are predominantly consisting of the
(111) planes [7,28]. Similar results were reported for
the evaporated films of Ag [29] and Au [29,30]. No
experimental result on the structure of the electrode
surfaces roughened by oxidation–reduction cycles is
available. It seems reasonable, however, to assume the
predominance of the (111) planes as in films. Then, it
would be desirable to carry out an ab initio calculation
for CH3S–metal cluster systems which can closely
mimic the adsorption of methanethiolate on (111)
metal surface planes. At the current level of calcula-
tion, we faced difficulty related to the convergence in
the SCF iteration when more than three metal atoms
were included. Hence, to gain an insight into the
adsorption of CH3SϪ at the three-fold sites on the
(111) planes, calculated results for CH3SM3 was
compared with those for CH3SM and CH3SM2. The
results in Table 2 show that the n(CS) frequency red-
shifts more as the number of the metal atoms, either

90
S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
Ag or Au, coordinated to methanethiolate increases.
This is in agreement with the experimental observa-
tion that this mode for methanethiolate adsorbed on
the metal surfaces appears at lower frequencies
compared to those for the respective salts. In parti-
cular, the red-shifts of this mode for CH3SAg3 from
the one for CH3SAg is comparable to the experimental
frequency difference between the silver salt and
SERS. As was mentioned earlier, the substantial
red-shifts of this mode in SERS compared to those
in salts are observed in all the thiolates investigated
in this work. This led us to conclude that the adsorp-
tion of the thiolates on the roughened silver electrode
surfaces occurs mainly on the multiple sites, the three-
fold hollw sites in particular.
In the case of Au, however, the experimental n(CS)
frequency difference between the salts and SERS is
generally smaller than that between CH3SAu and
CH3SAu3 in the calculated results. The n(CS)
frequencies on the Au electrode surfaces are invari-
ably higher than the corresponding ones on Ag for all
the thiolates investigated. In contrast, the calculated
n(CS) frequencies for CH3SAun are lower than the
corresponding values for CH3SAgn for n ˆ 1, 2, and
3. These results can be reconciled only by assuming
that the three-fold sites on the Au electrode surfaces
are not as much favored as those on Ag for the adsorp-
tion of thiolate. This may imply that adsorption of
CH3SϪ on the Au electrode surface occurs mostly
on the on-top or two-fold bridge sites.
In the electron energy loss spectroscopic study on
the adsorption of dimethyl disulfide on Au(111) [31],
the molecule was found to adsorb dissociatively with
the C–S bond of the thiolate tilted from the surface
normal. Contrarily, the perpendicular orientation of
the C–S bond of methanethiolate with respect to the
Ag(111) surface was observed by the sum frequency
generation spectroscopy [32]. Similar trend was
suggested for the adsorption of alkanethiolates with
long chains by the quantitative analysis of the infrared
spectral data [9–11]. According to the high level ab
initio study by Sellers et al. [19], the C–S surface
angles for methanethiolate adsorbed at the on-top
sites of Ag(111) and Au(111) planes are ϳ 105Њ,
while those at the hollow sites are 180Њ. Combining
these experimental and computational results, one
finds that the adsorption of methanethiolate occurs
mostly at the on-top sites on Au while at the hollow
sites on Ag. This is in agreement with our results from
the analysis of the spectral shift described earlier.
Also noteworthy is the fact that the relative intensity
of the n(CS) mode on the gold electrode is weaker
than that on the silver electrode (see Fig. 3).
According to the electromagnetic selection rule for
SERS proposed by Creighton [33] and by Moskovits
and Suh [34], vibrations with nonvanishing tensor
components along the surface normal display larger
enhancement in their Raman intensities than others.
This is in agreement with the present results.
However, it is well known that the surface enhance-
ment is also affected by other factors such as the
charge transfer [35].
According to the calculation by Sellers et al. [19],
adsorption at the hollow sites is energetically more
favorable that that at the on-top sites, slightly more
so on Au(111) than on Ag(111). Hence, it is difficult
to understand why multiple sites on the Au electrode
surfaces are less favored than those on the Ag elec-
trode. One can guess that it may have something to do
with the fact that the valence orbitals of atomic silver
are significantly more diffuse and extend higher above
the surface than those of gold. Nonetheless, it should
be recalled that even though the n(CS) frequencies of
all the thiolates on the Au surfaces are consistently
higher than the corresponding values on the Ag
surfaces, the frequency difference between the gold
salts and those on the Au surfaces are very diverse.
For example, rather small differences are observed for
methanethiol, ethanethiol, propanethiol (G), butha-
nethiol (G), and cyclohexanethiol, while substantially
larger differences are seen for propanethiol (T), butha-
nethiol (T), and 2-methyl-2-propanethiol. Hence, we
cannot generalize on the nature of the binding sites on
the Au surfaces. All that can be said is that the number
of atoms on the Au surface contributing to the chemi-
sorption of thiolates is smaller on the avearage than
those on the Ag surface.
It was well documented that the crystal structure of
the surface of the gold substrate does influence the
structure of the self-assembled alkanethiol mono-
layers. For instance, from a helium atom diffraction
study the alkyl chains were concluded to be more
tilted from the surface normal on a Au(110) surface
than on Au(111) [36]. Contrarily, the alkyl chains
were packed more densely on Au(100) than on
Au(111) and (110) surfaces [6,36,37]. In this respect,

S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
91
[7] C.A. Widrig, C.A. Alves, M.D. Porter, J. Am. Chem. Soc. 113
(1991) 2805.
[8] R. Heinz, J.P. Rabe, Langmuir 11 (1995) 506.
[9] M.D. Porter, T.B. Bright, D.L. Allara, C.E.D. Chidsey, J. Am.
Chem. Soc. 109 (1987) 3359.
the detailed structure of the roughened gold electrode
has to be clarified in the near future to understand the
shift of the C–S stretching frequencies of alkanethio-
lates on gold more precisely.
[10] M.M. Walczak, C. Chung, S.M. Stole, C.A. Widrig, M.D.
Porter, J. Am. Chem. Soc. 113 (1991) 2470.
[11] P.E. Laibinis, G.M. Tjatjopoulos, D.L. Allara, Y. Tao,
A.N. Parikh, R.G. Nuzzo, J. Am. Chem. Soc. 113 (1991)
7152.
[12] T.H. Joo, K. Kim, M.S. Kim, J. Phys. Chem. 90 (1986)
5816.
[13] C.K. Kwon, K. Kim, M.S. Kim, J. Mol. Struct. 197 (1989)
171.
[14] S.H. Cho, H.S. Han, D.-J. Jang, K. Kim, M.S. Kim, J. Phys.
Chem. 99 (1995) 10594.
[15] M.A. Bryant, J.E. Perberton, J. Am. Chem. Soc. 113 (1991)
8284.
[16] C.A. Szafranski, W. Tanner, P.E. Laibinis, R.L. Garrell,
Langmuir 14 (1998) 3570.
[17] S.B. Lee, K. Kim, M.S. Kim, W.S. Oh, Y.S. Lee, J. Mol.
Struct. 296 (1993) 5.
[18] K. Lee, S. Park, H. Kim, Bull. Kor. Chem. Soc. 16 (1995)
89.
[19] H. Sellers, A. Ulman, Y. Shnidman, J.E. Eilers, J. Am. Chem.
Soc. 115 (1993) 9389.
[20] T.H. Joo, M.S. Kim, K. Kim, J. Raman Spectrosc. 18 (1987)
57.
[21] R. Puddephatt, The Chemistry of Gold, Elsevier, Oxford,
1978.
5. Conclusion
Methanethiolate chemisorbs on the Ag or Au
surfaces mainly via its 2e and 3a1 molecular orbi-
tals. The 2e orbitals are virtually nonbonding with
respect to the C–S bond possessing only a slight
antibonding character, while the 3a1 orbital is
bonding. Hence, even though the 2e orbitals
contribute more to the chemisorption than the
3a1 orbital, the frequency shift of the n(CS)
mode is more affected by the electron donation
from the latter than from the former. In the case
of the Ag surface, adsorption of thiolates seems to
occur at the multiple sites, such as the three-fold
hollow sites. The nature of the adsorption sites on
the Au surfaces is less clear, being more or less
adsorbate dependent, even though the number of
the Au atoms interacting with thiolates are less
than that of the Ag atoms on the average.
[22] D.C. Harris, Quantitative Chemical Analysis, Freeman, New
York, 1995.
[23] P. Gao, D. Gosztola, L.H. Leung, M.J. Weaver, J. Electroanal.
Chem. 243 (1987) 211.
Acknowledgements
[24] M.W. Schmitt, K.K. Baldridge, J.A. Boatz, S.T. Elbert, M.S.
Gordon, J.H. Jensen, S. Koseki, N. Matsunaga, K.A. Nguyen,
S.J. Su, T.L. Windus, M. Dupuis, J.A. Montgomery, J.
Comput. Chem. 14 (1993) 1347. Original program assembled
by the staff of the NRCC: M. Dupuis, D. Spangler, J.J.
Wendoloski, National Resource for Computations in
Chemistry Software Catalog Program QG01, University of
California, Berkeley, CA, 1980.
This work was supported by the Korea Science
and Engineering Foundation through the Center
for Molecular Catalysis at Seoul National Univer-
sity (SNU) and by the Korea Research Foundation
through the Basic Science Research Institute at
SNU.
[25] W.J. Stevens, H. Basch, M. Krauss, P. Jasien, Can. J. Chem.
70 (1992) 612.
[26] A.G. Massey, N.R. Thompson, B.F.G. Johnson, R. Davis, The
Chemistry of Copper, Silver, and Gold, Pergamon, Oxford,
1973.
[27] S.H. Hong, U.Z. Olin, R. Hesse, Acta Chem. Scand. A 29
(1975) 583.
[28] V.N. Hallmark, S. Chiang, J.F. Rabolt, J.D. Swalen, R.
Wilson, J. Phys. Rev. Lett. 59 (1987) 2879.
[29] S. Buchholz, H. Fuchs, J.P. Rabe, J. Vac. Sci. Technol. B 9
(1991) 857.
References
[1] G.M. Whitesides, P.E. Laibinis, Langmuir 6 (1990) 87.
[2] A. Ulman, An Introduction to Ultrathin Organic Films from
Langmuir–Blodgett to Self-Assembly, Academic Press,
Boston, 1991.
[3] L.H. Dubois, R.G. Nuzzo, Annu. Rev. Phys. Chem. 43 (1992)
437.
[4] C.A. Alves, E.L. Smith, M.D. Porter, J. Am. Chem. Soc. 114
(1992) 1222.
[5] A. Ulman, Chem. Rev. 96 (1996) 1533.
[6] L.H. Dubois, B.R. Zegarski, R.G. Nuzzo, J. Chem. Phys. 98
(1993) 678.
[30] M.S. Zei, Y. Nakai, G. Lehmpfuhl, D.M. Kolb, J. Electroanal.
Chem. 150 (1983) 201.
[31] R.G. Nuzzo, B.R. Zegarski, L.H. Dubois, J. Am. Chem. Soc.
109 (1987) 733.

92
S.I. Cho et al. / Journal of Molecular Structure 479 (1999) 83–92
[32] A.L. Harris, L.H. Rothberg, L.H. Dubois, N.J. Levinos, L.
Dhar, Phys. Rev. Lett. 64 (1990) 2086.
[33] J.A. Creighton, Surf. Sci. 124 (1983) 209.
[34] M. Moskovits, J.S. Suh, J. Phys. Chem. 88 (1984) 5526.
[35] R.L. Birke, T. Lu, J.R. Lombardi, in: R. Varma, J.R. Selman
(Eds.), Techniques for Characterization of Electrodes and
Electrochemical Processes, Wiley, New York, 1991, p. 211.
[36] N. Camillone, C.E.D. Chidsey, G.-Y. Liu, G. Scoles, J. Chem.
Phys. 98 (1993) 4234.
[37] L. Strong, G.M. Whitesides, Langmuir 4 (1988) 546.