Subnanometric near-field Raman investigation in the vicinity of a metallic nanostructure.

Abstract

We present a near-field Raman investigation in the subnanometric vicinity of a metallic nanotip, where the tip-sample distance is precisely controlled by our newly developed time-gated illumination technique. Using this scheme on an isolated carbon nanotube, we have profiled the spatial decay of evanescent light. We also investigated extremely short-ranged chemical and mechanical interactions between the metal on the tip apex and the molecules of an adenine sample, which are observable only within the subnanometric vicinity of the tip. The results show a near-field Raman investigation with an accuracy of better than a few angstroms. Further, this shows strong promise for superhigh resolution in optical microscopy based on this technique.

Full text

Page 1

Subnanometric Near-Field Raman Investigation in the Vicinity of a Metallic Nanostructure
Taro Ichimura,1,2,*Shintaro Fujii,1Prabhat Verma,1,2,3,†Takaaki Yano,1,2Yasushi Inouye,2,3and Satoshi Kawata1,2,4
1Department of Applied Physics, Osaka University, Osaka 565-0871, Japan
2CREST, Japan Corporation of Science and Technology, Japan
3Graduate School of Frontier Biosciences, Osaka University, Osaka 565-0871, Japan
4RIKEN, Saitama 351-0198, Japan
(Received 13 November 2008; revised manuscript received 26 March 2009; published 4 May 2009)
We present a near-field Raman investigation in the subnanometric vicinity of a metallic nanotip, where
the tip-sample distance is precisely controlled by our newly developed time-gated illumination technique.
Using this scheme on an isolated carbon nanotube, we have profiled the spatial decay of evanescent light.
We also investigated extremely short-ranged chemical and mechanical interactions between the metal on
the tip apex and the molecules of an adenine sample, which are observable only within the subnanometric
vicinity of the tip. The results show a near-field Raman investigation with an accuracy of better than a few
angstroms. Further, this shows strong promise for superhigh resolution in optical microscopy based on this
technique.
DOI: 10.1103/PhysRevLett.102.186101 PACS numbers: 68.37.Uv, 73.20.Mf, 78.30.?j
Tip-enhanced Raman scattering (TERS) spectroscopy
[1–5] has been established as a powerful optical imaging
tool that allows one to optically investigate samples at the
nanometric scale. This technique can overcome the dif-
fraction limit of light, and hence a microscopy based on
TERS can provide subwavelength optical images far be-
yond the diffraction limit [6–8], which is attributed to the
highly confined evanescent field at the apex of a metal-
coated nanotip that is utilized as a probe in this technique.
The dominant mechanism for interaction between the tip
and the sample is electromagnetism (EM), with an inter-
action length of the order of a few to a few tens of nano-
meters. However, because of the presence of metal on the
tip, the chemical interaction between the sample molecules
and the metal starts to become significant, if the sample is
very close to the tip, particularly, at molecular distances
[9,10]. Further, if the distance between the sample and the
tip is reduced so much that the tip comes in physical
contact with the sample, then mechanical interaction be-
tweenthetipandthesamplealsostartstoplayanimportant
role [5,11]. Since the three interaction mechanisms have
different interaction lengths, it could be possible to study
them individually, if the distance between the tip and the
sample in a TERS experiment is precisely regulated with
accuracies of the order of subnanometer scale. This is an
extremely difficult task, because the nanotip that is usually
controlled by atomic force microscopy (AFM) operating
either in contact [12] or in tapping [13] mode, cannot be
maintained at a preselected fixed distance from the sample
during a long measurement. In this Letter, we demonstrate
how we can effectively control the tip-sample distance in
TERS experiments with extremely high accuracy, even
though the tip in our experiment oscillates under the
tapping-mode operation of AFM. We demonstrate that
the chemical as well as the mechanical interaction dis-
tinctly shows up in TERS spectra, when the tip is main-
tained very close to, or pressed against the sample. We
achieve this result by time gating the illumination, so that
the sample is selectively illuminated only when the tip is at
a certain distance from the sample during the repeated
cycles of its tapping oscillation. This technique allows us
to preselect any desired distance between the tip and the
sample, and then maintain that distance for as long expo-
sure as we want, by repeatedly illuminating the sample
only for that chosen tip-sample distance during every cycle
of tip oscillation. In the first part of this study, we utilized
an isolated single-walled carbon nanotube (SWNT) as a
sample to profile the distance dependency of EM interac-
tion. In the later part, we used a nanocrystal of adenine to
demonstrate clear signatures of chemical and mechanical
interactions on TERS spectra for precisely controlled sub-
nanometric tip-sample distances. We also show a one-
dimensional scan of an isolated SWNT to emphasize the
possibility of super high spatial resolution in an imaging
technique based on mechanical interaction in TERS.
Efforts for controlling distance between tip and sample
have received some interest in the near past for fluores-
cence [14,15] and for Raman scattering [16] measure-
ments. These experiments usually involve sheer-force
control of the AFM tip, which has poor stability and
accuracy of tip position. On the other hand, a time-gated
scheme with tapping-mode AFM control can have better
control over the tip position. We have recently shown that a
synchronous time gating of photon counting can effec-
tively provide a good control over tip-sample distance in
TERS measurements [17]. However, this scheme can only
analyze the optical signal strength, such as Raman or
fluorescence intensity. The chemical and mechanical inter-
actionsaffect thespectral shapesby inducingbotha shift in
frequency and a change of intensity of Raman modes.
Therefore it is essential to record whole spectrum, rather
than just the intensity of a Raman mode. We therefore
PRL 102, 186101 (2009)
PHYSICALREVIEW LETTERS
week ending
8 MAY 2009
0031-9007=09=102(18)=186101(4)186101-1
? 2009 The American Physical Society

Page 2

developed a new experimental scheme, where, instead of
the detector, we apply a synchronous time gating to the
illumination system and record entire spectrum by a multi-
channel detector. This allows us to record the entire spec-
trum for any desired tip-sample distance at extremely high
accuracy.
A schematic of our experimental setup is shown in
Fig. 1. Figure 1(a) illustrates that the tip-sample arrange-
ment is selectively illuminated only for a particular tip-
sample distance. The sinusoidal oscillation of the tip [18]
and the synchronized opening of the time gate are shown
by the upper and the lower curves, respectively, in
Fig. 1(b). One can preselect a desired tip-sample distance
and the accuracy in measurement by selecting certain
values of the time delay (?) and the temporal width (?),
respectively, of the time gate. This is done by feeding a
trigger signal extracted from the oscillation of the tip to a
pulse generator, which synchronously generates pulsed
signal withdesiredamplitudeandwidth.Thispulsedsignal
controls an optical shutter through an acousto-optic modu-
lator for stroboscopic illumination of the tip-sample ar-
rangement, as illustrated in Fig. 1(c).
We first discuss the tip-sample distance dependency of
the EM interaction, which is expected to follow the spatial
profile of the evanescent field in thevicinity of the tip apex.
We selected SWNTs as a sample in this study, which were
synthesized by the direct-injection pyrolytic-synthesis
method [19]. The sample was prepared by dispersing
SWNTs on a glass substrate, and then selecting an isolated
SWNT with length much longer than the focal spot of the
illumination. Raman scattering from the sample was ex-
citedwith ? ¼ 488 nmthroughan oil-immersionobjective
lens (NA ¼ 1:4) and the scattered signal was recorded by a
CCD detector after dispersing by a Raman spectrometer. A
commercially available silicon cantilever tip was coated
with a 28-nm-thick silver layer by thermal vapor deposi-
tion, which was positioned on the sample in the focal spot
and was controlled through the tapping-mode operation of
an AFM. The pulse generator was set in such a way that for
the tapping period of 8 ?s, the temporal width of the time
gate, ?, was 0:2 ?s. The resolution in tip-sample distance,
which depends on the absolute position of the tip, was
estimated to be about 3 A˚when the tip was closest to the
sample.The accuracyintheabsolutepositionofthetip was
about 1 A˚. Figure 2(a) displays some of the TERS spectra,
marked by A through D, recorded at specified tip-sample
distances, d, which is measured from the upper surface of
the SWNT. Spectrum A at d ¼ 0 nm represents a physical
contact between the tip and the sample, whereas spectrum
D at d ¼ 100 nm represents a far-field spectrum. The
prominent Raman peaks appearing in the spectral range
of 1550–1620 cm?1, are characteristic Raman modes of
SWNTs, known as the G band. The intensities of the peaks
in the G band increase drastically as the tip-sample dis-
tance decreases, which reflects the near-field enhancement
in TERS.
While we have shown only four spectra in Fig. 2(a), we
actually measured a large number of spectra for various
values of d. The intensity of the G-band signal was found
to be consistently increasing with decreasing tip-sample
distance, as shown by the data points in Fig. 2(b). In order
to numerically confirm the nature of field confinement at
the tip apex, we performed a simulation based on the finite-
difference time-domain (FDTD) formulation. The tip apex
in the simulation was chosen to be 30 nm in size, and it was
assumed that 30 nm length of an isolated infinitely long
SWNT with a diameter of 1 nm was illuminated with this
confined field. The simulation was performed by gradually
changing the tip-sample distance and Raman intensity
FIG. 1 (color online).
lectively illuminated only for a predecided tip-sample distance.
(b) The sinusoidal oscillation of the tip and the synchronized
opening of the time gate are illustrated. The tip-sample distance
and the resolution in the distance can be selected by choosing
certain values of ? and ?. (c) The complete system of tip-
enhanced Raman spectroscopy using a tapping-mode AFM and
acousto-optic modulator for time-gated illumination.
(a) The tip-sample arrangement is se-
1700 1600 15001400
Raman shift [cm-1]
d = 0.0 nm
d = 0.5 nm
d = 2.2 nm
d = 100.0 nm
(a)(b)
A
B
C
D
d [nm]
1/e = 2.8 nm
Raman intensity [arb. unit]
Raman intensity [arb. unit]
20
15
10
5
0
6040 20
0
FIG. 2 (color online).
at four different indicated tip-sample distances. (b) A plot of
integrated Raman intensity of the G-band versus the tip-sample
distance d. The solid curve represents a profile obtained from
FDTD simulation.
(a) Raman spectra of an isolated SWNT
PRL 102, 186101 (2009)
PHYSICALREVIEWLETTERS
week ending
8 MAY 2009
186101-2

Page 3

profile was obtained in the z direction. The simulated
profile, which shows an exponential decay, is plotted by
red curve in Fig. 2(b), and it shows great agreement with
the experimental results. The decay length, defined as the
value of d where the intensity drops by a factor 1=e, is a
parameter that quantifies the spatial confinement of light at
the tip apex and was found to be 2.8 nm, which confirms
strongly confined nature of the light field near the tip apex.
In this study, we experimentally observed the spatial
behavior of EM interaction between the tip and the sample.
However, a clear evidence of chemical or mechanical
interaction was not observed. Nevertheless, since this sam-
ple shows strong response to EM interaction, we selected
this sample for studying the behavior of EM interaction.
Next, in order to concentrate on the chemical and mechani-
cal interactions to complement the EM interaction, we
selected adenine molecule, because it has stronger affinity
with silver, causing a stronger chemical interaction. Also,
our experiments showed that it was easy to compress
adenine sample by slight push of the tip, making it easier
to see mechanical interaction even for a small amount of
tip-applied pressure.
Self-assembled adenine nanocrystals [20] were prepared
by casting ethanol solution of adenine on a glass substrate,
and an isolated nanocrystal was selected as a sample. All
experimental parameters were the same as those used for
the previous sample. The maximum tip-applied force on
adenine nanocrystal during the oscillation of the tip was
about 0.4 nN, which was estimated from experimental
parameters [21]. The adenine nanocrystal was estimated
to be compressed by 1.0 nm under this force. Since d is
measured from the upper surface of the sample, it takes a
negative value when the sample is compressed under tip-
applied pressure in the present experiment.
Figure 3(a) shows a data set of TERS spectra from
adenine nanocrystal for the indicated tip-sample distances.
The spectra are shown in the spectral range of the ring-
breathing mode (RBM) of adenine, which shows a promi-
nent peak at about 721 cm?1. Spectrum A was taken when
the sample was compressed under tip-applied force, spec-
trum B was taken when the tip was in contact with the
sample, but the average force applied by the tip on the
samplewas negligible, and spectrum C was taken when the
tip was very close to, but not in contact with the sample.
For a comparison, spectrum D was taken for the maximum
distance between the tip and the sample, which represents
far-field Raman scattering.
The far-fieldspectrum, shownbyD inFig.3(a),could be
well fitted with one Lorentzian function, whereas the spec-
tra A–C could be best fitted with two Lorentzian functions,
showing the presence of two underlaying Raman peaks.
The peak appearing in spectrum D at 721 cm?1was ob-
served in all spectra at the same frequency position. We
term this peak, which corresponds to the unperturbed RBM
of adenine, as !0. The other peak appeared at 731 cm?1in
spectra C and B, whereas it appeared at 736 cm?1in
spectrum A. We term this peak as !1. The dependencies
of frequency positions and intensities of both peaks !0
and !1on the tip-sample distance are plotted in Figs. 3(b)
and 3(c), respectively. Figure 3(c) also shows the increase
of integrated intensity (for combined !0and !1modes)
with decreasing tip-sample distance, which represents total
near-field enhancement of RBM.
Spectrum D in Fig. 3(a) represents Raman scattering
from adenine nanocrystal under no influence of the tip. As
the tip came close to the sample at a distance of 0.1 nm,
some of the adenine molecules started to chemically inter-
act with silver metal on the tip, and the RBM vibrational
frequency of those molecules shifted to 731 cm?1. This
shift of 10 cm?1is consistent with our earlier theoretical
prediction [9]. At the same time, there were many other
adenine molecules within the focal spot, which were not
close enough to the tip, and hence they continued to have
their RBM frequency at the original position of 721 cm?1.
As a result, spectrum C in Fig. 3(a) shows two Raman
peaks, !0and !1. When the tip-sample distance was
further reduced to d ¼ 0 in spectrum B, more adenine
molecules came under chemical interaction with the metal
on the tip, resulting in slight increase in the intensity of !1
FIG. 3 (color online).
crystal at indicated tip-sample distances. The peaks marked by
!0and !1represent the unperturbed RBM and the frequency-
shifted RBM, respectively. Deconvoluted Lorentzian peaks for
each spectrum are also shown by the dotted curves. (b) Tip-
sample distance dependence of Raman shift, and (c) of the peak
intensity, of the two Raman modes !0and !1. The dashed lines
indicate the surface of the sample (d ¼ 0). An integrated inten-
sity plot is also shown in (c), which represents total enhancement
of the RBM. (d) A schematic illustration of 1D scan of an iso-
lated SWNT along a direction perpendicular to its axis. (e) The
corresponding TERS intensity increases when the tip is exactly
above the SWNT, depicting a spatial resolution of 3 nm.
(a) Raman spectra of an adenine nano-
PRL 102, 186101 (2009)
PHYSICAL REVIEWLETTERS
week ending
8 MAY 2009
186101-3

Page 4

mode and slight decrease in the intensity of !0mode, as
also shown in Fig. 3(c). Further, when the adenine nano-
crystal was pushed by the tip to a value of d ¼ ?1:0 nm,
the corresponding Raman spectrum, presented as A in
Fig. 3(a), showed significant changes in both frequency
position and peak intensity for the mode !1. These drastic
changes could be explained by an additional effect of
mechanical interaction due to the uniaxial pressure applied
by the tip. It has been shown previously that tip-applied
pressure can significantly shift Raman modes [5,11]. As
the tip compressed the sample, the contact area between
the tip and the sample increased, resulting in an increase in
the number of sample molecules in contact with the tip.
Because of the contact, these molecules underwent chemi-
cal as well as mechanical interactions at the same time
[22]. The chemical interaction provided a shift of 10 cm?1
to !1, as also seen in spectrum B. The mechanical inter-
action induced a further shift of 5 cm?1to the RBM of the
same molecules, resulting in a total shift of 15 cm?1to !1
in spectrum A. The additional shift of 5 cm?1due to tip-
applied pressure is consistent with expected amount to
pressure-induced shift in the RBM of adenine [9,10]. The
increased number of sample molecules in contact with the
tip is one of the dominant reasons for an increase of the
intensity of !1mode in spectrum A. The other possibility
is due to modified resonance condition under tip-applied
pressure [11].
Invoking desired pressure effect by precisely controlling
the tip position can be of tremendous benefit in super high
resolution imaging, because the pressure effect is much
morelocalizedthan theenhancedEMfieldinthex-yplane.
While the EM field extends to a size comparable to the tip
apex, the tip-applied pressure extends only to the area of
physicalcontact.Owingtothecurvatureofthetipapex,the
contact area in present experiment can be smaller than
1 nm, even if the tip apex is about 30 nm. Therefore, an
optical imaging based on precisely controlled tip-applied
pressure can provide spatial resolution much better than
what is currently obtained in TERS imaging. In order to
support our claim, we performed a one-dimentional scan of
an isolated SWNTin the direction perpendicular to its axis,
while keeping the tip-sample distance within the range
where the pressure effects in TERS could be observed. In
this scan, the tip was moved at steps of 3 nm, and TERS
was measured at 4 consecutive positions as the tip crosses
over the SWNT. A schematic of this scan process is shown
in Fig. 3(d). For the entire scan, the EM enhancement was
almost constant, as the total scan area was much smaller
than the extent of the enhanced field. However, as shown in
Fig. 3(e), the TERS intensity noticeably increases at posi-
tion 3, in comparison with positions 1, 2 and 4. This
increase of TERS intensity is associated with the pressure
effect [11], which is highly localized in x-y plane. The
increase of TERS intensity in Fig. 3(e) shows that the
SWNT exists at position 3, which is observed with a high
spatial resolution of at least 3 nm. The results show strong
promises for super high spatial resolution in imaging tech-
nique based on precisely controlled tip-sample distance in
TERS.
In conclusion, we have developed a new time-gated
technique for stroboscopic illumination in TERS experi-
ments, which effectivelyallows usto have a precise control
over measuring TERS spectrum for a preselected tip-
sample distance. Using this scheme, we have demonstrated
an exponentially decaying nature of evanescent light near
the apex of a metal-coated tip. A precise control of the tip-
sample distance with subnanometric accuracy allowed us,
for the first time, to experimentally observe the influences
of extremely short-ranged chemical and mechanical inter-
actions between the tip and the sample in a TERS mea-
surement. This study opens doors for direct observations,
and hence for further studies, of chemical and mechanical
interactions in TERS, which range typically within sub-
nanometer scale. Further, it shows strong promise for
TERS imaging with super high spatial resolution.
The authors thank Dr. Takeshi Saito of AIST, Japan,
for supplying the SWNT samples used in this study. This
work was supported by a grant from the Japan Science and
Technology Agencyunder a Core Research forEvolutional
Science and Technology (CREST) project ‘‘Plasmonic
Scanning Analytical Microscopy.’’
*ichimura@ap.eng.osaka-u.ac.jp
†verma@ap.eng.osaka-u.ac.jp
[1] Y. Inouye et al., Proc. SPIE Int. Soc. Opt. Eng. 40, 3791
(1999).
[2] R.M. Sto ¨ckle et al., Chem. Phys. Lett. 318, 131 (2000).
[3] N. Hayazawa et al., Opt. Commun. 183, 333 (2000).
[4] B. Pettinger et al., Phys. Rev. Lett. 92, 096101 (2004).
[5] P. Verma et al., Phys. Rev. B 73, 045416 (2006).
[6] A. Hartschuh et al., Phys. Rev. Lett. 90, 095503 (2003).
[7] T. Ichimura et al., Phys. Rev. Lett. 92, 220801 (2004).
[8] T. Yano et al., Appl. Phys. Lett. 88, 093125 (2006).
[9] H. Watanabe et al., Phys. Rev. B 69, 155418 (2004).
[10] N. Hayazawa et al., J. Chem. Phys. 125, 244706 (2006).
[11] T. Yano, Y. Inouye, and S. Kawata, Nano Lett. 6, 1269
(2006).
[12] W. Zhang et al., J. Phys. Chem. C 111, 1733 (2007).
[13] A. Rasmussen and V. Deckert, J. Raman Spectrosc. 37,
311 (2006).
[14] F.M. Huang, F. Festy, and D. Richards, Appl. Phys. Lett.
87, 183101 (2005).
[15] P. Anger, P. Bharadwaj, and L. Novotny, Phys. Rev. Lett.
96, 113002 (2006).
[16] B. Pettinger et al., Phys. Rev. B 76, 113409 (2007).
[17] T. Yano et al., Appl. Phys. Lett. 91, 121101 (2007).
[18] J.P. Cleveland et al., Appl. Phys. Lett. 72, 2613 (1998).
[19] T. Saito et al., J. Phys. Chem. B 109, 10647 (2005).
[20] B. Giese and D. McNaughton, J. Phys. Chem. B 106, 101
(2002).
[21] Q. Zhong et al., Surf. Sci. Lett. 290, L688 (1993).
[22] T. Ichimura et al., J. Phys. Chem. C 111, 9460 (2007).
PRL 102, 186101 (2009)
PHYSICAL REVIEW LETTERS
week ending
8 MAY 2009
186101-4