PHYSICAL REVIEW E 87, 020401(R) (2013)
Activated vibrational modes and Fermi resonance in tip-enhanced Raman spectroscopy
Mengtao Sun,1Yurui Fang,1Zhenyu Zhang,2,3and Hongxing Xu1,4,5,*
1Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, P.O. Box 603-146,
Beijing 100190, People’s Republic of China
2ICQD, Hefei National Laboratory at the Microscale, University of Science and Technology of China, Hefei, People’s Republic of China
3Department of Physics and Astronomy, University of Tennessee, Knoxville, Tennessee 37996-1200, USA
4Division of Solid State Physics, Lund University, Lund 22100, Sweden
5The Center of Nanoscience and Technology and School of Physics and Technology, Wuhan University, Wuhan 430072,
People’s Republic of China
(Received 23 August 2011; revised manuscript received 23 January 2013; published 13 February 2013)
Using p-aminothiophenol (PATP) molecules on a gold substrate and high-vacuum tip-enhanced Raman
spectroscopy (HV-TERS), we show that the vibrational spectra of these molecules are distinctly different
from those in typical surface-enhanced Raman spectroscopy. Detailed first-principles calculations help to
assign the Raman peaks in the TERS measurements as Raman-active and IR-active vibrational modes of
to DMAB under the TERS setup. The activation of the IR-active modes is due to enhanced electromagnetic field
gradient effects within the gap region of the highly asymmetric tip-surface geometry. Fermi resonances are also
observed in HV-TERS. These findings help to broaden the versatility of TERS as a promising technique for
ultrasensitive molecular spectroscopy.
DOI: 10.1103/PhysRevE.87.020401PACS number(s): 33.20.Fb, 68.37.Uv, 68.43.Pq, 73.20.Mf
First demonstrated by St¨ ockle et al., Hayazawa et al., and
Anderson in 2000 [1–3], tip-enhanced Raman spectroscopy
(TERS) is a high-sensitivity optical analytical technique,
with high spatial resolution beyond the diffraction limit of
light. In TERS, a sharp metal tip is used to create a “hot
site” to excite localized surface plasmons and, consequently,
enhance the electromagnetic field and Raman signals in the
vicinity of the tip apex [1–9]. The tip can be moved in three
dimensions to control the position of the hot site and the
the gap between the tip and the substrate. Therefore, TERS
has the inherent advantage of overcoming one of the most
severe restrictions in the application of surface-enhanced
Raman scattering (SERS), which usually requires roughness
of metal surfaces or aggregations of metal nanoparticles to
create hot sites that are difficult to control. TERS may solve a
wide variety of problems in high-vacuum (HV) single-crystal
surface science, electrochemistry, heterogeneous catalysis,
As a compelling and challenging example, the appearance
of normally unseen Raman modes in the spectrum of p-
aminothiophenol (PATP) adsorbed on metal surfaces has been
a long-standing issue in SERS. Earlier explanations often
modes of PATP due to charge transfer between the molecules
and the metal substrates [10–13]. Recently, dimerization of
PATP to dimercaptoazobenzene (DMAB) has been suggested
to explain these unseen Raman modes [14–18]. Both the-
oretical simulations [14,15] and experimental observations
[15,16] offer credible support to the explanation based on the
*Corresponding author: email@example.com
dimerization mechanism; nevertheless, controversies remain
regarding the dimerization explanation due to the broadening
and relatively few Raman modes in the SERS spectra [19–21].
It has been suggested that the gradient effect associated
with the strongly enhanced electric field at a rough metal
surface [22–24] could activate IR-active modes, which are
usually Raman inactive. If the IR-active modes of DMAB
are observed together with the Raman-active modes in SERS
studies, it is expected that the dimerization picture will show
more convincing spectroscopic evidence.
In this Rapid Communication, we exploit the advantage of
TERS to investigate the long-standing issue of the appearance
of new Raman modes of adsorbed PATP molecules, which
are typically unseen in normal Raman spectroscopy. We
observe many more and narrower Raman peaks in our high-
vacuum TERS (HV-TERS) system than in previous SERS
studies. By comparing these experimental observations with
our theoretical simulations within density functional theory,
we can assign almost all the Raman peaks as Raman-active
symmetric modes and IR-active asymmetric modes of the
DMAB molecule. The fully activated and strongly enhanced
IR-active modes as observed in our HV-TERS data give more
PATP via dimerization into the putative product DMAB. The
gradient effect associated with the greatly enhanced electric
field within the nanogap region between the sharp gold tip
and gold film can activate the IR-active modes of the DMAB
molecule to be Raman active. The huge electromagnetic
enhancement in the nanogap region further results in distinct
Fermi resonances, the strong coupling of a fundamental mode
and an overtone of a different mode or a combinational mode,
to split the corresponding fundamental Raman modes.
The schematic diagram of the HV-TERS system is shown
in Fig. 1. It consists of a homemade scanning tunneling
microscope (STM) in a high-vacuum chamber, a Raman
1539-3755/2013/87(2)/020401(5)©2013 American Physical Society
MENGTAO SUN, YURUI FANG, ZHENYU ZHANG, AND HONGXING XU PHYSICAL REVIEW E 87, 020401(R) (2013)
FIG. 1. (Color online) Schematic diagram of the home-built HV-
spectrometer combined with side illumination of 632.8 nm
He-Ne laser light with an angle of 60◦for Raman mea-
surements, and three-dimensional piezo stages for the tip
and sample manipulations. The pressure in the chamber is
∼10−7Pa. A gold tip with a ∼50 nm radius was fabricated by
The substrate was prepared by evaporating a 100 nm gold film
to a newly cleared mica film under high vacuum. The film
was immersed in a 1 × 10−5M PATP ethanol solution for
24 h, then washed with ethanol for 10 min to guarantee that
there was only one monolayer of PATP molecules adsorbed on
the gold film. Then, immediately, the sample was put into the
high-vacuum chamber. To get a good signal-to-noise ratio, the
TERS signals were collected with an acquisition time of 60 s
and accumulated 20 times for each spectrum.
We measured TERS of PATP adsorbed on the Au film at
different bias voltages and currents and found the optimal
conditions to be ±1 V (bias voltage on the sample) and
1 nA (current) in our HV-TERS system. Two series of typical
measured spectra using the experimental conditions described
above for different biases are shown in Figs. 2(a) and 2(b).
The fluctuations of these spectra are small, and the spectra
are stable and can be readily repeated experimentally under
these conditions. The profiles of the spectra obtained at
+1 V and −1 V are quite similar, indicating that there is
not much influence from the polarity of the bias voltage on
the TERS measurement here. For comparison, the typical
SERS spectrum of PATP in Au sol and the normal Raman
scattering (NRS) spectrum of PATP powder [Fig. 2(c)] were
also measured with a Renishaw inVia Raman system and
excited with light at 632.8 nm. The molecular structures of
DMAB and PATP are also shown in the insets of Fig. 2(c).
Comparing Figs. 2(a) and 2(b) with Fig. 2(c), it is found
that the TERS peaks of PATP adsorbed on the Au film are
significantly different from the SERS and NRS peaks of
PATP. Many more and narrower Raman peaks are observed
in Figs. 2(a) and 2(b) for the HV-TERS study than in Fig. 2(c)
for the SERS and NRS studies. It is worthwhile to note that
our previous report showed that the SERS peaks of PATP
adsorbed on a bare Au film are very similar to the NRS
of PATP , whereas the SERS peaks of PATP on Au
colloidal nanoparticles are different from the corresponding
D M A B
P A TP
Intensity (arb. units)
FIG. 2. Spectra of DMAB under experimental conditions of 1 nA
positions with the HV-TERS system. (c) SERS peaks of DMAB in
Au sol and the normal Raman scattering spectrum of PATP.
NRS measurements. The latter case has been attributed to
plasmon-assisted chemical reaction of PATP dimerization into
DMAB, but that is not the case for the bare Au film where the
in TERS could also suggest PATP dimerization due to a strong
local electromagnetic enhancement within the nanogap region
of the tip-surface geometry. Moreover, the polarity of the bias
voltage does not influence the TERS profiles in Figs. 2(a) and
2(b), which may indicate that the detected molecule should
be structurally symmetric, offering additional evidence for the
surfaces symmetrically (the Au tip and the Au film) with two
thiol groups. Nevertheless, the difference between TERS and
SERS spectra remains.
To interpret the above spectral phenomena correctly, we
choose two spectra at different biases for vibrational mode
assignments, as shown in Figs. 3(a) and 3(b). Furthermore,
the Raman spectra of DMAB were calculated within density
functional theory  using the GAUSSIAN 09 suite  with
the PW91PW91 functional (Perdew–Wang’s 1991 exchange
functional plus Perdew–Wang’s 1991s correlation functional)
, a 6-31G(d) basis set for C, N, S, and H, and a LANL2DZ
Au, where a Au5-DMAB-Au5junction was used to simulate
the Raman spectra, mimicking the TERS setup. The IR
spectra of DMAB in the Au5-DMAB-Au5junction were also
simulated with the same method. Based on the simulated
Raman spectrum in Fig. 3(c), all the Raman-active symmetric
ACTIVATED VIBRATIONAL MODES AND FERMI ...
PHYSICAL REVIEW E 87, 020401(R) (2013)
1100 1200 1300 1400 1500 1600
IR Intensity (10 km/mole)
Raman Intenstiy (104A4/ amu)
TERS intensity (arb. units)
FIG. 3. (Coloronline)TERSpeaksofDMABunderexperimental
conditions of 1 nA current and (a) −1 V and (b) +1 V bias voltage.
(c) Simulated Raman and (d) IR spectra of DMAB, where the wave
number is scaled by 1.014.
ag vibrational modes in Figs. 3(a) and 3(b) can be assigned.
Interestingly, most of the remaining Raman peaks in Figs. 3(a)
and 3(b) can be assigned as IR-active asymmetric bu modes
previously shown that the gradient-field effect can activate IR-
activemodes inSERSduetomolecular quadrupole transitions
when the molecules are placed near metal surfaces [22,23].
Such gradient-field effects were also observed in near-field
optical microscopy Raman (NSOM-Raman) . Similarly,
the observation of the full activation of IR-active modes can
be caused by very high field gradients within the nanogap of
a highly asymmetric geometry, consisting of a sharp metal tip
enormous electromagnetic field enhancement in the nanogap
region makes this phenomenon of activation of IR-active
modes more readily observable.
Next, we compare the simulated Raman spectrum in
Fig. 3(c) and the SERS peaks in Fig. 2(c). It is found that six
strong peaks in the experimental and simulated results agree
well with each other without scaling. We also note that in the
simulated Raman spectrum of DMAB in Fig. 3(c), the wave
numbers have been scaled by an insignificant factor of 1.014
(or blueshifted by ∼1%) for closer comparison with the TERS
data shown in Figs. 3(a) and 3(b). Such blueshifts could easily
Intensity (arb. units)
Intensity (arb. units)
FIG. 4. (Color online) (a) Experimentally observed FR and
simulated TERS peaks of DMAB for the Raman-active symmetric
ag13and IR-active asymmetric bu13modes. (b) and (c) Calculated
vibrational modes and the combinational modes for FR, where the
wave number is scaled by 1.014.
be caused by weak external perturbations, such as tip, current,
and voltage in the HV-TERS system.
Figure 4(a) shows a zoom-in spectrum of Fig. 3(a),
where each of the symmetric and asymmetric C-N stretching
modes, assigned as the Raman-active ag13mode and IR-active
asymmetric bu13mode, respectively, splits into two Raman
peaks. The splitting of the Raman peaks results from Fermi
resonance (FR). In FR, an overtone of a different fundamental
mode or a combination mode can appear in the vibrational
spectra by gaining spectral weight from a fundamental mode
. FRs are frequently found in IR or Raman spectra in
symmetric triatomic molecules such as CO2and CS2[30–32]
but not in a TERS study. When the molecules are placed in the
nanogap region, external perturbations from greatly enhanced
electromagnetic fields, high field gradients, and the Au tip
could cause such distinct FRs.
For a FR, the split vibrational energies under the perturba-
tion can be written as [30–32]
2(EA+ EB) ±1
(EA− EB)2+ 4φ2,
where EAand EBare the vibrational energies of the funda-
mental mode and an overtone of a different fundamental mode
(or a combination mode) before splitting and φ is the FR
coupling coefficient, which describes the coupling strength
MENGTAO SUN, YURUI FANG, ZHENYU ZHANG, AND HONGXING XUPHYSICAL REVIEW E 87, 020401(R) (2013)
of the fundamental vibrational mode and the combinational
coefficient is larger, the spectral energy splitting between the
fundamental vibrational mode and the combinational mode
(or the overtone mode) is also larger. In addition, when the
unperturbed energies EAand EBin Eq. (1) are closer, the FR
coupling coefficient becomes larger. For our case in Fig. 4(a),
the unperturbed energies EA and EB for the Raman-active
symmetric ag13 or the IR-active asymmetric bu13 can be
where I+and I−are the Raman intensities of the split peaks.
The difference in energy between the perturbed levels in the
presence and absence of FR can be obtained with ?E±=
E+− E−and ?EAB= |EA− EB|, respectively.
In Fig. 4(a), we obtain I+(ag13) ≈ I−(ag13) and I+(bu13) ≈
2I−(bu13), ?E±(ag13) = 20.4 cm−1, and ?E±(bu13) =
14.6 cm−1. For the Raman-active symmetric ag13 mode,
according to Eqs. (2) and (3), we obtain EA= EB=E++E−
i.e., ?EAB≈ 0 cm−1, and the FR coupling coefficient can
be calculated by using Eq. (1) as φ(ag13) =1
10.2 cm−1. As illustrated in Figs. 4(b) and 4(c) with dif-
ferent vibrational modes marked as short vertical lines, the
fundamental Raman-active symmetric ag13mode frequency at
1213 cm−1is very close to the combinational mode frequency
of ag6 at 727 cm−1and bg6 at 485 cm−1. with ?EAB=
E(ag13) −?E(bg6) + E(ag6)?≈ 1 cm−1. The combinational
For the IR-active asymmetric bu13 mode, we obtain
to Eqs. (2) and (3), and I+(bu13) ≈1
We can then obtain ?EAB(bu13) =1
mode bg × ag = bg is asymmetric.
2I−(bu13) as in Fig. 4(a).
3?E±(bu13) = 4.9 cm−1
and φ(bu13) =
Figs. 4(b) and 4(c), the calculated IR-active asymmetric bu13
combinational mode of bu5at 634 cm−1and bu6at 640 cm−1.
(4.9 and 7 cm−1) is about 2 cm−1. The symmetry of the
combinational mode is bu × bu = ag is symmetric.
In summary, by using the HV-TERS system, we have
provided evidence for the conversion of PATP to DMAB via
dimerization, possibly catalyzed by the presence of the TERS
tip on top of the gold film. The observation in HV-TERS
of the IR-active bands associated with DMAB molecules
provides further experimental evidence for the dimerization
mechanism. The activation of these extra modes is primarily
attributed to enhanced electromagnetic field gradient effects
within the gap region of the tip-surface geometry. Moreover,
strongly enhanced plasmon responses in the nanogap region
further result in distinct Fermi resonance to split the corre-
the capacity of TERS as an analytical tool for ultrasensitive
3?E±(bu13) = 6.88 cm−1. As illustrated in
This work was supported by the National Basic Re-
search Project of China (Grants No. 2012YQ12006005,
2007CB936804), the National Natural Science Foundation of
China (Grants No. 90923003, No. 10874233, No. 10874234,
No. 20703064, No. 11034006, and No. 11227407), the
Knowledge Innovation Project (Grant No. KJCX2-EW-W04)
of CAS, the U.S. National Science Foundation (Grant No.
0906025), the U.S. Department of Energy, Basic Energy
Sciences, Materials Sciences and Engineering Division, and
the Nanometer Structure Consortium at Lund University,
 R. M. St¨ ockle, Y. D. Suh, V. Deckert, and R. Zenobi, Chem.
Phys. Lett. 318, 131 (2000).
 N. Hayazawa, Y. Inouye, Z. Sekkat, and S. Kawata, Opt.
Commun. 183, 333 (2000).
 M. S. Anderson, Appl. Phys. Lett. 76, 3130 (2000).
 B. Pettinger, B. Ren, G. Picardi, R. Schuster, and G. Ertl, Phys.
Rev. Lett. 92, 096101 (2004).
 W. Zhang, X. Cui, B. S. Yeo, T. Schmid, C. Hafner, and R.
Zenobi, Nano Lett. 7, 1401 (2007).
 E. Bailo and V. Deckert, Chem. Soc. Rev. 37, 921 (2008).
 J. Steidtner and B. Pettinger, Phys. Rev. Lett. 100, 236101
 M. T. Sun, Z. L. Zhang, H. R. Zheng, and H. X. Xu, Sci. Rep. 2,
 N. Jiang, E. T. Foley, J. M. Klingsporn, M. D. Sonntag, N. A.
and R. P. Van Duyne, Nano Lett. 12, 5061 (2012).
 M. Osawa, N. Matruda, K. Yoshii, and I. Uchida, J. Phys. Chem.
98, 12702 (1994).
 A. Campion and P. Kambhampati, Chem. Soc. Rev. 27,
 Q. Zhou, X. Li, Q. Fan, X. Zhang, and J. Zheng, Angew. Chem.,
Int. Ed. 45, 3970 (2006).
 T. Shegai, A. Vaskevich, I. Rubinstein, and G. Haran, J. Am.
Chem. Soc. 131, 14390 (2009).
 D. Y. Wu, X. M. Liu, Y. F. Huang, B. Ren, X. Xu, and Z. Q.
Tian, J. Phys. Chem. C 113, 18212 (2009).
 Y. R. Fang, Y. Z. Li, H. X. Xu, and M. T. Sun, Langmuir 26,
 Y. F. Huang, H. P. Zhu, G. K. Liu, D. Y. Wu, B. Ren, and Z. Q.
Tian, J. Am. Chem. Soc. 132, 9244 (2010).
 M. T. Sun and H. X. Xu, Small 8, 2777 (2012).
 Y. Z. Huang, Y. R. Fang, Z. L. Yang, and M. T. Sun, J. Phys.
Chem. C 114, 18263 (2010).
 W.-H. ParkandZ.H.
 K. Kim, D. Shin, H. B. Lee, and K. S. Shin, Chem. Commun.
47, 2020 (2011).
ACTIVATED VIBRATIONAL MODES AND FERMI ...
PHYSICAL REVIEW E 87, 020401(R) (2013)
 W. Ji, N. Spegazzini, Y. Kitahama, Y. Chen, B. Zhao, and
Y. Ozaki, J. Phys. Chem. Lett. 3, 3204 (2012).
 M. Moskovits and D. P. DiLella, J. Chem. Phys. 73, 6068
 E. J. Ayars, H. D. Hallen, and C. L. Jahncke, Phys. Rev. Lett.
85, 4180 (2000).
 B. Ren, G. Picardi, and B. Pettinger, Rev. Sci. Instrum. 75, 837
 P. Hohenberg and W. Kohn, Phys. Rev. 136, B864 (1964).
 M. J. Frisch et al., GAUSSIAN 09, revision A.02, Gaussian, Inc.,
Wallingford, CT, 2009.
 J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R.
Pederson, D. J. Singh, and C. Fiolhais, Phys. Rev. B 46, 6671
 P. J. Hay and W. R. Wadt, J. Chem. Phys. 82, 270 (1985).
 E. Fermi, Z. Phys. 71, 250 (1931).
 R. A. Nyquist, H. A. Fouchea, G. A. Hoffman, and D. L. Hasha,
Appl. Spectrsoc. 45, 860 (1991).
 K. D. Bier and H. J. Jodi, J. Chem. Phys. 86 4406 (1987).