Content uploaded by Susumu Yanagisawa
Author content
All content in this area was uploaded by Susumu Yanagisawa on Jun 15, 2015
Content may be subject to copyright.
Role of Molecular Orbitals Near the Fermi Level in the Excitation of Vibrational Modes
of a Single Molecule at a Scanning Tunneling Microscope Junction
Michiaki Ohara,
1,2
Yousoo Kim,
1
Susumu Yanagisawa,
3
Yoshitada Morikawa,
3,4
and Maki Kawai
1,2
1
Surface Chemistry Laboratory, RIKEN, Saitama 351-0198, Japan
2
Department of Advanced Materials Science, The University of Tokyo, Chiba 277-8561, Japan
3
The Institute of Scientific and Industrial Research, Osaka University, Osaka 567-0047, Japan
4
Research Institute for Computational Sciences, National Institute of Advanced Industrial Science and Technology,
Ibaraki 305-8568, Japan
(Received 1 February 2007; published 2 April 2008)
Inelastically tunneled electrons from the tip of a scanning tunneling microscope were used to induce S-
S bond dissociation of a CH3S2and lateral hopping of a CH3Son Cu(111) at 4.7 K. Both experimental
results and theoretical calculations confirm that the excitation mechanism of the vibrationally induced
chemistry reflects the projected density of states of molecular orbitals that appear near the Fermi level as a
result of the rehybridization of the orbitals between the adsorbed molecules and the substrate metal atoms.
DOI: 10.1103/PhysRevLett.100.136104 PACS numbers: 68.37.Ef, 68.35.Ja, 68.43.Fg, 68.43.Pq
Mode-selective chemistry, whereby a specific chemical
bond of a molecule is excited to select a desired reaction
pathway, has been widely investigated in the gas phase
[1,2]. The scanning tunneling microscope (STM) provides
a way to achieve mode-selective chemistry of single mole-
cules adsorbed on surfaces because inelastically tunneled
electrons from the STM tip can excite molecular vibrations
that lead to a variety of dynamic motions and chemical
reactions [3–16]. Molecular vibrations excited through
inelastic electron tunneling processes can be accompanied
by changes in the differential conductance of a molecule.
Inelastic electron tunneling spectroscopy with STM (STM-
IETS) [17], which records the second derivative of the
tunneling current (I) with respect to bias voltage (V), has
proven useful for detecting vibrational signals of individual
molecules. However, it does not always reveal all the
vibrational modes in the spectra, nor is it applicable to
mobile or reactive molecules [4–6,11,15,17]. As an alter-
native to the STM-IETS, action spectroscopy, which mea-
sure the response of vibrationally mediated molecular
motion to applied bias voltage, has been used for obtaining
vibrational spectra of mobile molecules. Sainoo et al. have
shown that action spectroscopy can even detect vibrational
modes that are not visible using STM-IETS [11].
Persson and Baratoff [18] developed a theory of STM-
IETS using an adsorbate-induced resonance model, and
before the first experiment by Stipe et al. [3], they pre-
dicted that the STM-IET spectrum would exhibit a peak-
shaped signal at the particular bias voltage where the
vibrational excitation of a molecule takes place.
Similarly, for the action spectrum obtained from the reac-
tion induced by vibrational excitation, Ueba and Persson
[19] have shown that the second derivative of the reaction
rate with respect to the bias voltage provides direct access
to the vibrational density of states. Moreover, the width of
this action spectrum displays the intrinsic vibrational
broadening of a single adsorbed molecule on the surface.
Although action spectra offer rich information of this
kind, the precise mechanism by which a particular vibra-
tional mode induces molecular reactions by inelastic tun-
neling electrons has not yet been experimentally examined.
Many theoretical attempts to explain the tunneling
electron-vibration coupling [20] in a molecule have been
based on a resonance model, in which an incident electron
becomes temporarily captured by molecular orbitals (MO)
localized in a resonant state resulting in excitation of
molecular vibrations. Thus, a detailed knowledge of the
MOs of the adsorbate is necessary for understanding the
underlying mechanism of electron-vibration coupling and
for predicting which vibrational mode is actually excited
by tunneling electrons.
This Letter reports on quantitative investigations of the
vibrationally induced bond dissociation of an isolated
CH3S2molecule and the lateral hopping of an isolated
CH3Smolecule adsorbed on the Cu(111) surface with a
low-temperature STM. Based on the experimental findings
and on density functional theory (DFT) calculations, we
discuss how the spatial distribution of MOs of an adsorbate
at resonant states can affect electron-vibration coupling
and thereby affect the resulting molecular motions and
reactions.
All STM measurements were carried out at 4.7 K [21,22]
with a low-temperature STM (LT-STM, Omicron GmbH)
in an ultrahigh-vacuum chamber ( <31011 Torr).
Typical conditions for obtaining STM images were
Vsample 20 mV and Itunnel 0:2nA. The clean
Cu(111) surface was exposed to CH3S2vapor at a tem-
perature below 50 K. After the STM image was taken, the
STM tip was positioned over the center of a target mole-
cule. The feedback loop was then turned off, and the
tunneling electrons with the defined tunneling current
were injected into the molecule. During electron injection,
this tunneling current stays constant except when the target
molecule reacts, for example, by bond dissociation or
PRL 100, 136104 (2008) PHYSICAL REVIEW LETTERS week ending
4 APRIL 2008
0031-9007=08=100(13)=136104(4) 136104-1 ©2008 The American Physical Society
molecular hopping, in which case the tunneling current
suddenly decreases. In the present study, we calculated the
mean time required for bond dissociation (before molecu-
lar hopping actually occurred) for each applied bias volt-
age over 10 trial events with the STM tip in the same
position. Multiplying the value of the current (I) and the
mean time taken for the reaction (T) gives the number of
electrons necessary for the dissociation and for the hop-
ping, from which the reaction yield (R) is determined. The
reaction probability was plotted against applied bias volt-
age to produce an action spectrum. DFT calculations were
carried out using the program package DMOL3in the
Materials Studio of Accelrys Inc. [23,24] for optimizing
the adsorption system, and the program code STATE
(Simulation Tool for Atom TEchnology) [25–27] for the
projected density of states (PDOS) analysis and the draw-
ing of MOs for an isolated CH3S2and CH3Son Cu(111).
Individual CH3S2molecules on the Cu(111) surface
appear as elliptic protrusions in STM images, as shown in
Fig. 1(a). The intersection points of the mesh in Fig. 1(a)
indicate the center positions of the Cu atoms on the surface
[21]. By injecting tunneling electrons into the target mole-
cule, we find that a single CH3S2molecule is broken into
two identical ball-shaped protrusions, implying S-S bond
dissociation to form two CH3Smolecules [Fig. 1(b)]. For
the adsorption site of CH3S, fcc and hcp hollow sites are
equally eligible [28].
Action spectra for dissociation of CH3S2and CD3S2
are shown in Fig. 1(c). Each spectrum shows one clear
threshold voltage at 357.5 mV for CH3S2and at 275 mV
for CD3S2. The threshold voltage at 357.5 mV corre-
sponds to the reported vibrational excitation energy of a
C-H stretching mode (C-H) of the CH3group in the
molecule. In high-resolution electron energy loss spectros-
copy (HREELS), the vibrational frequencies of the C-H
mode are observed at 365 meV for monolayers of CH3S2
on Au(111) and at 366 meV for those on Cu(100) [29,30].
Moreover, the shift of the threshold voltage from 357.5 to
275 mV reasonably corresponds to the isotope shift of
C-Hto C-D. Note that the small increase in the
reaction yield (-䉬- and --) clearly appears at around
410 mV for CH3S2and at 330 mV for CD3S2, respec-
tively [31]. In the negative bias voltage region, however,
the reaction yield was 1:61012 and 0:91012 for
Vsample 800 mV and 600 mV, respectively. This
negligibly small dissociation yield in the negative bias
voltage region suggests that the reaction yield is strongly
influenced by whether the electron was injected to the
molecule or extracted from it.
Excited molecular vibrations can cause dynamic pro-
cesses by overcoming the potential barrier along the reac-
tion coordinate (RC) of a specific adsorbate motion. It has
been shown that the vibrational mode along the RC can be
excited not only directly but also indirectly through cou-
pling to the higher frequency (HF) vibrational mode ex-
cited by tunneling electrons [9]. In the present study, the
thresholds observed in the action spectra as a sharp peak at
357.5 meV due to C-Hand as a shoulder at 410 meV
due to C-HS-Sreveal that S-S bond dissociation
in a CH3S2molecule is induced by excitation of two HF
modes, not by direct excitation of the S-Smode, which
is the RC mode for S-S bond dissociation. This is the first
observation of an indirect pathway for intramolecular bond
dissociation on a surface attained by means of tunneling
electrons. So far, only a direct pathway, resulting from the
direct excitation of a specific RC mode, has been reported,
in experiments involving O-O bond dissociation of O2on
Pt(111) [12] and C-H bond dissociation of trans-2-butene
on Pd(110) [15].
Figure 2(a) shows an STM image of two CH3Smole-
cules obtained by breaking the S-S bond of a CH3S2
molecule on Cu(111). In previous work, we reported that
the injection of tunneling electrons can induce individual
CH3Smolecules to hop randomly between the fcc and hcp
hollow sites of Cu(111) [22], as shown in Figs. 2(b)–2(d).
An action spectrum for the hopping of CH3Sis shown in
Fig. 2(e). In the spectrum, threshold voltages are observed
at both 85 mV, the absolute value of which corresponds
to the vibrational excitation energy of the C-S stretching
mode (C-S). It is important to note that the reaction
yield was identical for both bias polarities, which is very
different from what occurs in CH3S2dissociation.
According to DFT calculations, the vibrational excitation
energy of C-Sis 85 meV for CS3Son a Au substrate and
81 meV for CS3Son a Cu substrate [26,30]. Experi-
mentally, the C-Smode of various alkylthiolates has
been observed at similar energies in HREELS: for ex-
ample, 88 meV for methylthiolate on Au(111) and
81 meV for butylthiolate and hexylthioate on Au(111)
[32]. Since the frustrated translation and/or rotation modes
are the RC mode for the motion [7,9], as for the dissocia-
FIG. 1. Topographic STM images of CH3S2molecules on Cu
(111) before (a) and after (b) injection of tunneling electrons.
(c) Action spectra for the S-S bond dissociation of CH3S2and
CD3S2molecules. Initial current was set to 4 nA. The lower
traces with ‘‘-䉬-’’ and ‘‘--’’ represent the slopes of the reaction
for CH3S2and CD3S2molecules, respectively. In both spec-
tra, each data point consists of 10 trial events, and thus to obtain
these spectra, a total of 130 molecules for CH3S2and of 110
molecules for CD3S2respectively were dissociated by tunnel-
ing electrons, with the STM tip condition kept constant.
PRL 100, 136104 (2008) PHYSICAL REVIEW LETTERS week ending
4 APRIL 2008
136104-2
tion of the S-S bond in a CH3S2molecule, the hopping of
CH3Sis evidently also the result of an indirect mechanism
of RC mode excitation via anharmonic coupling between
HF C-Sand RC modes.
As mentioned above, the S-S bond dissociation of
CH3S2is induced by vibrational excitation of the
C-Hmode, whereas lateral hopping of CH3Sis not,
although both molecules involve C-H bonds of the methyl
(CH3) group. Sainoo et al. have pointed out that vibrational
modes observed in action spectra can be explained by
means of a resonant model of electron-vibration coupling
[11]. When a molecule is chemisorbed onto the metal
surface, it is distorted by molecule-metal bond formation,
and the rehybridized MO consists of several fundamental
MOs that appear around the Fermi level (Ef). Thus, to
address the applicability of this resonant model to our
system, we calculated the projected density of states
(PDOS) and the spatial distribution of the MOs for each
molecular adsorbate, as shown in Figs. 3(a) –3(d).
Figure 3(a) shows that the lowest unoccupied molecular
orbital (LUMO) of CH3S2is located just above Ef, and it
also contributes to states near Ef. In addition, the small
contributions from the LUMO 1and LUMO 2states
are broadly distributed over the energy region shown. The
reaction yield for dissociation showed distinctly different
dependence on polarity, where the contribution of the
LUMO state apparently dominates, obscuring the rela-
tively small contributions from LUMO 1and LUMO
2as well as from the HOMO state which is distributed
along Cu-S bond. In the case of CH3S, the HOMO is
located just below Efand, in addition to the LUMO and
LUMO 1, it also contributes to states near Ef, as shown
in Fig. 3(c). The reaction yield for hopping is identical for
the applied bias voltages of both polarities, which parallels
the homogeneous distribution of the DOS across the Efof
the LUMO and/or LUMO 1, while that of the HOMO
contributes much more in the occupied state. The spatial
distribution of individual MOs for CH3S2and CH3Sis
depicted in Fig. 3(b) and 3(d), respectively. As shown in
Fig. 3(b), the LUMO of CH3S2is clearly localized at both
the S-S and C-H bonds in the molecule, suggesting that a
resonantly captured electron in the LUMO will influence
the S-S and C-H bond. However, in the case of CH3S
[Fig. 3(d)], there is no contribution of the MOs to the C-
H bond in the molecule. This explains why CH3S2shows
aC-Hsignal in the action spectrum, whereas CH3Sdoes
not. Figure 3(d) also reveals that the LUMO and LUMO
1are rather localized at the C-S bond in CH3S, which
supports the experimental observation of a C-Ssignal in
the action spectrum of the molecule. It is important to note
that the LUMO 1and LUMO 2of CH3S2are like-
wise localized at the C-S bond in the molecule, as shown in
Fig. 3(b). One would therefore expect that the C-Smode
also would be detected in the action spectrum of CH3S2.
However, in the present study, this was not the case.
Especially in the positive bias region, the cross-section
with the LUMO state dominates, and as a result, the con-
FIG. 3 (color online). (a) PDOS of an isolated CH3S2mole-
cule on Cu(111). (b) The spatial distributions of LUMO 2,
LUMO 1, LUMO, and HOMO for an isolated CH3S2mole-
cule on Cu(111). MOs on C-H bond in CH3S2molecule are
indicated by arrows. (c) PDOS of an isolated CH3Smolecule on
Cu(111). (d) The spatial distributions of LUMO 1, LUMO,
and HOMO for an isolated CH3Smolecule on Cu(111). H, C, S,
and Cu atoms are drawn in blue, yellow, green, and gray,
respectively.
FIG. 2. (a) –(d) Sequential STM images of the hopping of a
CH3Smolecule on Cu (111). (e) Action spectrum for the hop-
ping of CH3Smolecules in both positive and negative bias.
Initial current was set to 4 nA. The lower trace with ‘‘-䉬-’’
represents the slope of the hopping for CH3Smolecules in both
positive and negative bias. In the spectrum, each data point
consists of 10 trial events; thus, to obtain the spectrum in the
positive bias region, a total of 200 CH3Smolecules were
counted, with the STM tip maintained in a constant state. A
similar number of trials was undertaken for the negative bias
region.
PRL 100, 136104 (2008) PHYSICAL REVIEW LETTERS week ending
4 APRIL 2008
136104-3
tribution of LUMO 1and LUMO 2diminishes pro-
portionally. Thus, detailed interpretation of the spatial
distribution of individual MOs can successfully explain
our experimental action spectra, where the C-Hmode
is active for S-S bond dissociation of CH3S2and the
C-Smode is active for lateral hopping of CH3Sby
injecting inelastically tunneled electrons from the tip of
an STM.
In conclusion, a combination of quantitative STM ex-
periments and theoretical calculations has been used to
investigate the S-S bond dissociation of an isolated
CH3S2molecule and the lateral hopping of an isolated
CH3Smolecule on Cu(111) in terms of excitation of mo-
lecular vibrations induced by inelastically tunneled elec-
trons. Our experiments reveal that vibrational modes are
resonantly excited through a temporal occupation of the
MOs by the tunneling electrons and that the molecular
vibrations are selectively excited depending on the spatial
distribution and the population of the MOs near Efby
hybridization with the substrate metal.
Our present finding that molecular vibrations can be
selectively excited depending on the spatial distribution
of the MOs near Efoffers a way to systematically realize
mode-selective chemistry of individual molecules ad-
sorbed on surfaces, and it also may be applied to character-
ize molecules bridging two electrodes [33 –35].
The present work was supported in part by the Grant-in-
Aid for Scientific Research on Priority Areas, ‘‘Electron
Transport Through a Linked Molecule in Nano-Scale’’
(Grant No. 17069006), from the Ministry of Education,
Culture, Sports, Science and Technology, and ‘‘Control
and Application of Nano-Structural Materials for
Advanced Data Processing and Communications,’’
CREST program from Japan Science and Technology cor-
poration (JST). We are grateful for the computational
resources of the RIKEN Super Combined Cluster (RSCC).
[1] A. Sinha, R. L. Vander Wal, and F. F. Crim, J. Chem. Phys.
92, 401 (1990).
[2] P. R. Fleming, M. Li, and T. R. Rizzo, J. Chem. Phys. 94,
2425 (1991).
[3] B. C. Stipe, M. A. Rezaei, and W. Ho, Science 279, 1907
(1998).
[4] B. C. Stipe, M. A. Rezaei, and W. Ho, Phys. Rev. Lett. 81,
1263 (1998).
[5] J. Gaudioso, H. J. Lee, and W. Ho, J. Am. Chem. Soc. 121,
8479 (1999).
[6] W. Ho, J. Chem. Phys. 117, 11033 (2002), and references
therein.
[7] J. I. Pascual et al., Nature (London) 423, 525 (2003).
[8] Y. Kim, T. Komeda, and M. Kawai, Jpn. J. Appl. Phys. 41,
4924 (2002).
[9] T. Komeda et al., Science 295, 2055 (2002).
[10] C. Matsumoto et al., Surf. Sci. 587, 19 (2005).
[11] Y. Sainoo et al., Phys. Rev. Lett. 95, 246102 (2005).
[12] B. C. Stipe et al., Phys. Rev. Lett. 78, 4410 (1997).
[13] S.-W. Hla, L. Bartels, G. Meyer, and K.-H. Rieder, Phys.
Rev. Lett. 85, 2777 (2000).
[14] Y. Kim, T. Komeda, and M. Kawai, Surf. Sci. 502,7
(2002).
[15] Y. Kim, T. Komeda, and M. Kawai, Phys. Rev. Lett. 89,
126104 (2002).
[16] T. Komeda et al., J. Chem. Phys. 120, 5347 (2004).
[17] B. C. Stipe, M. A. Rezaei, and W. Ho, Science 280, 1732
(1998).
[18] B. N. J. Persson and A. Baratoff, Phys. Rev. Lett. 59, 339
(1987).
[19] H. Ueba and B. N. J. Persson, Phys. Rev. B 75, 041403(R)
(2007).
[20] H. Ueba, Surf. Rev. Lett. 10, 771 (2003), and references
therein.
[21] M. Ohara, Y. Kim, and M. Kawai, Langmuir 21, 4779
(2005).
[22] M. Ohara, Y. Kim, and M. Kawai, Chem. Phys. Lett. 426,
357 (2006).
[23] B. Delley, J. Chem. Phys. 92, 508 (1990).
[24] B. Delley, J. Chem. Phys. 113, 7756 (2000).
[25] Y. Morikawa, Phys. Rev. B 51, 14802 (1995).
[26] T. Hayashi, Y. Morikawa, and H. Nozoye, J. Chem. Phys.
114, 7615 (2001).
[27] S. Yanagisawa and Y. Morikawa, Chem. Phys. Lett. 420,
523 (2006).
[28] M. Ohara, Y. Kim, and M. Kawai, Jpn. J. Appl. Phys. 45,
2022 (2006).
[29] R. G. Nuzzo, B. R. Zegarski, and L. H. Dubois, J. Am.
Chem. Soc. 109, 733 (1987).
[30] B. A. Sexton and G. L. Nyberg, Surf. Sci. 165, 251 (1986).
[31] Second thresholds at 410 mV for CH3S2and at 330 mV
for CD3S2do not match with any of the vibrational
normal modes of CH3S2and CD3S2, respectively.
These higher voltages may correspond to a combination
mode of C-Hand S-S stretching [S-S] modes of
CH3S2. This is the first observation of a combination
mode related to bond dissociation. According to DFT
calculations, the vibrational excitation energies of the
S-Smode of isolated CH3S2and CD3S2molecules
on Cu(111) are 53 and 51 meV, respectively, which
suggests the higher threshold corresponds to the vibra-
tional excitation energy of the C-HSSmode.
[32] H. S. Kato et al., J. Phys. Chem. B 106, 9655 (2002).
[33] R. H. M. Smit et al., Nature (London) 419, 906 (2002).
[34] W. H. A. Thijssen, D. Djukic, A. F. Otte, R. H. Bremmer,
and J. M. van Ruitenbeek, Phys. Rev. Lett. 97, 226806
(2006).
[35] M. Kiguchi, R. Stadler, I. S. Kristensen, D. Djukic, and
J. M. van Ruitenbeek, Phys. Rev. Lett. 98, 146802 (2007).
PRL 100, 136104 (2008) PHYSICAL REVIEW LETTERS week ending
4 APRIL 2008
136104-4