th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
Development of Total Internal Reflection Raman Imaging for
Non-intrusive Quantitative Visualization of Near-wall Concentration
Reiko Kuriyama1,*, Tetsuro Tateishi1, Yohei Sato1
1: Dept. of System Design Engineering, Faculty of Science and Technology, Keio University, Japan
* correspondent author: [email protected]
Abstract The present study proposes a total internal reflection (TIR) Raman imaging technique for non-intrusive and
quantitative visualization of near-wall concentration distributions in micro- and nanofluidic devices. The measurement
principle relies on the chemical specificity of spontaneous Raman scattering and surface-selective excitation with an
evanescent wave which decays exponentially from the surface. A two-prism-based illumination system was employed
to generate the evanescent wave at a glass-solution interface. The Raman scattering arising from near-wall molecules
was two-dimensionally captured by an EM-CCD camera through an optical bandpass filter. Since the intensity of the
Raman signal is proportional to the number of molecules being illuminated, the Raman scattering images can be
converted into concentration distributions via calibration data. In this study, mixture solutions composed of H2O and
D2O were used as sample liquids focusing on their distinct Raman signals arising from O-H and O-D stretching
vibrations. According to the TIR Raman spectra of these solutions, two bandpass filters were selected for Raman
imaging in order to visualize the concentration of each species. Prior to the measurements, a calibration experiment was
carried out with each filter under uniform concentration distribution. For compensating the inhomogeneity of the
evanescent wave illumination, raw intensities were normalized by the reference intensity value at 50 mol% D2O
concentration. The calibration curve showed a linear relationship between the normalized intensity and the H2O or D2O
concentration. Applying the calibration result, a near-wall concentration measurement was performed in the mixing
flow field in a T-shaped microchannel. The lateral spatial resolution was 1.6 × 1.6 µm2 and the total acquisition time
was 30 seconds. The concentration distribution of each species was visualized in the evanescent wave spot positioned at
the junction of the channel where pure H2O and D2O merged. These results demonstrate that TIR Raman imaging offers
the potential to visualize near-wall molecular distributions quantitatively.
1. Introduction
Recent advances in microfluidic systems promise efficient approaches for miniaturizing and integrating
various chemical processes involved in biological analysis and chemical engineering, which has led to the
development of micro-total analysis systems or lab-on-a-chip devices (Arora et al. 2010). In such
miniaturized systems, interfacial phenomena have a noticeable effect on the liquid’s flow structure and
chemical properties due to the large surface-to-volume ratio. For example, an ionic layer called electric
double layer (Probstein 1994), which is formed at liquid-solid interfaces, produces electrokinetic flow and
affects the solution composition.
For further development of micro- and nanofluidic devices, many experimental efforts to probe these
surface-related events have been reported to date. Total internal reflection fluorescence (TIRF) microscopy,
which utilizes an evanescent wave to excite fluorophores only in the vicinity of the surface area, is often
used as a well-established diagnostic tool (Axelrod 2001). Using TIRF microscopy, Kazoe and Sato (2003)
developed a nanoscale laser induced fluorescence (nano-LIF) technique to measure the two-dimensional
distribution of the zeta potential which is a dominant parameter of electroosmotic flow. The measurement
principle takes advantage of the fact that fluorescent ions are distributed by electrostatic forces related with
the wall zeta potential. By applying the nano-LIF technique, color mapping of the zeta potential distribution
was successfully obtained at the channel wall with surface modification pattern (Kazoe et al. 2009). For
near-wall flow velocity measurement, Zettner and Yoda (2003) first demonstrated nano-particle image
velocimetry (nano-PIV), in which evanescent wave illumination was used for tracer-based velocimetry.
Nano-PIV has been applied to various flow fields to study fluid mechanics in interfacial regions, including
steady electroosmotic flows in a fused silica microchannel (Sadr et al. 2004) and transient electrokinetically
driven flow with non-uniform wall zeta-potential (Kazoe et al. 2010). While these fluorescence-based
methodologies offer sufficient sensitivity and high spatiotemporal resolution for the investigation of various
-1-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
near-wall phenomena, they have an inherent disadvantage of using extrinsic fluorescent labels which can
potentially affect the measurement results.
Different approaches based on Raman spectroscopy can be a powerful tool for surface-selective and
chemically-specific measurement. Surface-enhanced Raman spectroscopy (SERS) is a widely-used
technique, taking advantage of the local enhancement of the electromagnetic field at a nano-structured metal
surface due to excitation of surface-plasmon resonances (Kudelski 2009). Since SERS produces highly
increased Raman signal from molecules in contact with or in the proximity of the surface, it has been applied
to various fields of studies such as catalysis (Kim et al. 2010), biophysical chemistry (Hu et al. 2007) and
molecule detection in microfluidic sensors (Piorek et al. 2007). However the enhancement effect can be
achieved only with appropriately designed SERS-active metal surface; it makes the technique less versatile
or impractical in some applications.
Another methodology, total internal reflection (TIR) Raman spectroscopy, realizes non-intrusive
measurement without the need for any external markers or resonators by utilizing spontaneous Raman
scattering excited by an evanescent wave (Woods and Bain 2012). TIR Raman spectroscopy has long been
overshadowed by SERS since it was first demonstrated by Ikeshoji et al. in 1973, but modern
instrumentation makes it a practical alternative in applications where SERS is undesirable (Woods and Bain,
2012). Nickolov et al. (1993) investigated the water structure adjacent to a hydrophobic surface by
measuring TIR Raman spectra of OH stretching vibrational band of water. Tyrode et al. (2008) evaluated the
adsorbed amount of surfactants on a hydrophilic silica surface from the integrated intensities of TIR Raman
scattering. Although these studies have proved the effectiveness of TIR Raman spectroscopy, they have been
limited to single-point detection through a spectrometer.
The objective of the present study is to develop a non-intrusive and planar visualization technique of
near-wall molecular distributions. Total internal reflection Raman imaging is proposed by combined use of
spontaneous Raman imaging (Takahashi et al. 2012, Rinke et al. 2012) and evanescent wave illumination
generated at solid-liquid interface. Using this technique, Raman scattering from molecules in the vicinity of a
microchannel wall was two-dimensionally detected by a camera and converted into a concentration
distribution. H2O and D2O mixtures were utilized as sample liquids and a near-wall concentration map of
each species was visualized in the mixing flow in a T-shaped microchannel.
2. Measurement Principle
2.1 Evanescent Wave Illumination
When light passes from one medium (with higher refractive index, n1) to another (with lower refractive
index, n2), some is reflected back into the first medium and the other is transmitted into the second medium.
If the incident angle, θ i is smaller than the critical angle θ c, most of the light propagates into the lower index
medium at a refractive angle θ r which is given by Snell’s law. For supercritical condition (θ i > θ c), all of the
light is reflected, i.e., it undergoes total internal reflection. Even in this case, however, the electric field
penetrates through the interface and propagates parallel to the surface (in Figure 1). This field is termed
evanescent wave and its intensity, Ieva, decays exponentially with the perpendicular distance z from the interface, as expressed by
⎛ z ⎞
I eva ( z ) = I eva ( 0 ) exp ⎜ − ⎟ ,
⎜ zp ⎟
⎝
⎠
(1)
where
zp =
λ
4π n12sin 2θi − n2 2
(2)
.
The characteristic length zp is called the penetration depth and the parameter λ is the wavelength of the
incident light in vacuum. Since zp is generally in the order of λ or smaller except for supercritical θ i → θ c
-2-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
(where zp → ∞), the evanescent wave can selectively excite the molecules in the vicinity of the interface
without exciting the region farther from the interface (Axelrod 2001).
(b)
(a)
Evanescent wave
Penetration depth
n2 (< n1)
n1
(∼100 nm)
Z
Zp
0
θi
I0/e
I0 I
Reflected
Incident
Fig. 1 (a) Generation of evanescent wave and (b) its intensity distribution
2.2 Spontaneous Raman Scattering
When monochromatic light is incident on a sample, it is partially transmitted, absorbed and scattered.
Most part of the scattered light has the same frequency as the incident light (ν 0) due to the elastic process
called Rayleigh scattering. A small fraction of the scattered light (about 10-6 of the incident beam) occurs
however, as a result of inelastic collisions of photons and molecules, and thus is observed as a frequencyshifted emission (ν 0 ± ν i). This inelastic scattering event is known as Raman scattering. Raman shift (ν i) is
equal to the vibrational or rotational frequency of a molecule and its intensity has dependence with the
number of molecules being excited. Therefore chemically-specific quantitative analysis can be realized by
measuring Raman scattering intensity at the characteristic Raman shift.
Figure 2 shows the typical Raman spectra from bulk H2O/D2O mixtures. These mixtures present distinct
spectral shapes depending on their composition. The two broad Raman bands with peak intensities at 2500
cm-1 and 3420 cm-1 can be observed, which originate from O-D and O-H stretching vibrations, respectively.
In the present study, these H2O/D2O mixtures were used as sample liquids to demonstrate the ability of TIR
2Dsolutions
Graph 2 containing multiple components.
Raman imaging to probe the species concentrations in
Raman intensity
[a.u.]
Y Data
6000
6000
D2O concentration
100 mol% O-D stretching
4000
4000
75 mol%
50 mol%
25 mol%
0 mol%
O-H stretching
2000
2000
00
1500
1500
2000
2000
2500
2500
3000
3000
3500
3500
4000
4000
Data [cm-1]
RamanX shift
Col 1Raman
vs Col 2 spectra of bulk H2O/D2O mixtures
Fig. 2 Spontaneous
Col
Col
Col
Col
Col
1 vs
1 vs
1 vs
1 vs
1 vs
Col
Col
Col
Col
Col
3
4
5
6
7
2.3 Total Internal Reflection Raman Imaging
For non-intrusive visualization of near-wall concentration, TIR Raman imaging technique was proposed
for capturing Raman images from liquid molecules in contact with a solid surface. Figure 3 illustrates the
concept of the technique. The use of an evanescent wave enables the surface-selective excitation of
spontaneous Raman process at the solid-liquid interface. The Raman signal arising from near-wall liquid
-3-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
molecules is spectrally filtered by an optical filter and two-dimensionally captured by a camera. Since this
technique employs direct imaging of Raman scattering in the measurement area, there is no need for
traversing the light source or the sample. The resultant image is a distribution of the Raman scattering
intensity integrated over the transmittance wavelength of the filter, as shown in Figure 3. In order to ensure
the detection of only the desired Raman signal, the optical filter should be selected to transmit the
wavelength range that corresponds to the characteristic Raman signal of intended species. The expression for
the measured Raman intensity for the i-species, Ii [photons s-1 m-2], is
⎛ dσ ⎞
I i = k×I 0 × Ω × ⎜
⎟ × N i ,
⎝ dΩ ⎠i
(3)
where I0 [photons s-1 m-2] is the excitation laser intensity, Ω [sr] is the solid angle of signal detection,
dσ /dΩ [cm2 molecule-1 sr-1] is the absolute differential Raman cross-section and Ni is the number of ispecies molecules being illuminated. The constant k is introduced to express influencing parameters such as
transmission of the optical components and collection efficiency of the detector. The concentration of ispecies (Ci) can be obtained by exploiting the linear dependence of the Raman intensity (Ii) on the number of
molecules (Ni). Since the Raman intensity also depends on the excitation intensity (Eq. (3)), the measured
Raman intensity is normalized by that at the reference concentration (Ii_ref) as follows:
I ratio =
Ii
I i_ref
=
dσ
k×I 0 × Ω × ( dΩ
)i × Ni
dσ
k×I 0 × Ω × ( dΩ
)i × Ni_ref
(4)
= f (Ci ) .
By determining this Raman intensity ratio, the influence of the inhomogeneity of the evanescent wave
intensity can be cancelled out.
Evanescent wave
illumination
Raman intensity distribution
(2D Raman image)
Y
X
X
Calibration data
Raman scattering
Concentration
Filter
Near-wall concentration
distribution
Y
Camera
Intensity
Y
X
Fig. 3 Schematic concept of the near-wall concentration measurement by TIR Raman imaging
3. Measurement System for TIR Raman Imaging
3.1 Experimental Apparatus
The experimental setup is illustrated in Figure 4. A two-prism-based illumination system (Kazoe and
Sato 2007) was employed to generate an evanescent wave at the glass-solution interface in a microchannel.
This configuration has two advantages compared to the popular objective-based system (e.g., Axelrod 2001);
the first is a larger field of view and the second is wide selectivity of the channel substrate.
In the present study, a continuous-wave Nd:YVO4 laser (Coherent Inc., Verdi-V6, λ = 532 nm) was used
for an excitation light source. A convex lens was inserted in the laser light path in order to focus the beam
and to achieve the desired size of the evanescent wave spot. A microchannel fabricated from
poly(dimethylsiloxane) (PDMS) and a silica glass slide was positioned on the prisms with index matching
immersion oil. The focused laser beam was introduced into the silica glass through a prism and it traveled
-4-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
undergoing a certain number of total internal reflections with the incident angle of 75°. Afterwards the beam
was thrown out from the prism in the opposite side. Evanescent wave spots were generated at the glasssolution interface and the molecules only in the vicinity of the interface were excited. The Raman scattering
was collected through an objective lens (Nikon Corp., 10×, NA = 0.25), a dichroic mirror (> 555 nm) and an
optical bandpass filter, and imaged by an EM-CCD camera (Hamamatsu Photonics K. K., C9100-13, 512 ×
512 pixels, 16 bits). The selection of the bandpass filter will be described in detail below.
Mirror
Lens
Mirror
Microchannel
Evanescent wave spot
PDMS
Silica glass
75°
Prism
Objective lens
CW laser
Prism
(10 ×, NA = 0.25)
(532 nm, 0.43 W)
Dichroic mirror
(> 555 nm)
Bandpass filter
Raman image
(603−622 nm for D2O)
(641−658 nm for H2O)
Z
Y
EM-CCD camera
X
(512 × 512 pixels, 16 bits)
Fig. 4 Schematic of the experimental setup utilizing two-prism-based system
3.2 Microchannel
A major concern for detection of TIR Raman scattering is fluorescence and Raman signal arising from
channel substrates where TIR of the laser occurs. Since these undesirable signals often dominate the
measurement results, the substrate material should be carefully selected. This study focuses on the Raman
scattering with relatively large Raman shift (between 2000−4000 cm-1), thus silica glass which presents
strong Raman peak mainly below 1300 cm-1 (Tyrode et al. 2008) was used for the bottom wall of a
microchannel. The channel structure was molded in a PDMS layer by soft lithography (Xia and Whitesides
1998) and enclosed by a 1 mm-thick and 50 mm-diameter silica glass slide by utilizing self-adherence of
PDMS. A T-shaped microchannel with a height of 50 µm (Figure 5) was fabricated for a calibration
experiment and concentration visualization.
Top view
Evanescent wave spots
Major axis: 260 µm
Minor axis: 100 µm
Outlet
600 µm
Laser
300 µm
Inlet A
Inlet B
Cross-sectional view
PDMS
Silica glass
50 µm
(1.0 mm thick)
Fig. 5 Top and cross-sectional views of a T-shaped microchannel
3.3 Solution
-5-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
In the present experiments, five different mixture solutions composed of H2O and D2O were used as
samples. The properties of these solutions are listed in Table 1. The penetration depth of the evanescent
wave was calculated to be around 90 nm from Eq. (2), by substituting the laser wavelength (λ) of 532 nm,
incident angle (θ i) of 75°, refractive index of silica glass (n1) of 1.461 and that of the solution (n2) measured
by a refractometer (Atago Co., Ltd, PAL-RI).
Table 1 Properties of H2O/D2O mixtures used for the calibration experiment
A
B
C
D
H2O concentration [mol%]
100
75
50
25
0
D2O concentration [mol%]
0
25
50
75
100
Refractive index [-]
1.332
Penetration depth [nm]
91.4
1.331
90.8
1.330
90.3
1.329
89.7
E
1.328
89.2
.
3.4 TIR Raman Spectroscopy for Filter Selection
Prior to the experiment of TIR Raman imaging, spectroscopic measurements were performed in order to
choose the appropriate optical bandpass filters for concentration imaging, because it is important to separate
the desired Raman signal from noise. Figure 6(a) shows the TIR Raman spectra from the H2O/D2O mixtures
(in Table 1) and that from an empty channel (i.e., filled with ambient air). These were obtained by using the
aforementioned illumination system, a grating spectrometer (Solar TII Ltd., 400 groove/mm) and the EMCCD camera. The horizontal axis in the upper side is the wavelength of the scattered light excited at 532 nm.
The total acquisition time per spectrum was 100 s and the spectral resolution was less than 0.57 nm (about 15
cm-1). The nominal laser power was set to be 0.50 W. For noise reduction purposes, the measured spectra
were smoothed with a 11-points, 3rd-order Savitzky-Golay filter (Savitzky and Golay 1964).
For all the spectra in Figure 6(a), broad Raman bands are observed at 1600 cm-1 and 3700 cm-1, which
were not observed in the spectra from bulk mixtures (Figure 2). Since these bands are observed for the empty
channel (shown as red line), they are attributed to silica glass being used as a channel substrate. In addition
to these strong Raman bands, the relatively weak signals arising from O-H and O-D stretching vibrations can
be observed around 3300 cm-1 and 2500 cm-1, respectively. Although their band shapes are distorted by the
signal from the substrate, it can be seen that their intensities increase slightly with the concentration of the
corresponding species. In order to remove the Raman signal from silica glass and show the signal from
mixtures more clearly, difference spectra are calculated for the five solutions (Figure 6(b)) by subtracting the
reference spectrum of the empty channel. The difference spectra exhibit the characteristic band shapes of OH and O-D stretching modes as in bulk spectra. The intensity change with mixture composition is also
confirmed from Figure 6(b). According to the above spectral measurement result, two optical filters were
selected for concentration imaging; one is for H2O with transmittance wavelength of 641−658 nm (Semrock
Inc., FF01-650/13-25) and the other is for D2O with transmittance wavelength of 603−622 nm (Semrock
Inc., FF01-615/20-25). The transmittance wavelength ranges are indicated by arrows and colored areas in
Figure 6(b). The figure shows that their transmittance wavelengths correspond to the spectral region of the
O-H and O-D stretching Raman bands and there is little overlap between them.
Finally, the integrated intensity of the TIR spectrum in the transmittance range of each filter was
calculated for the solutions and plotted against the concentration in Figure 6(c). It is obvious that the
integrated intensity for each filter linearly increased with the concentration of corresponding species, which
is explained by Eq. (3). These results indicate that concentration measurement of desired species can be
realized by TIR Raman imaging by utilizing the appropriately-selected filters.
-6-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
2D Graph 2
Wavelength [nm]
(a)
Raman intensity [a.u.]
3000
3000
580
2000
2000
1500
640
660
Empty channel
(Silica glass)
1000
1000
500
00
1500
1500
2000
2000
Col
Col
Col
Col
Col
Col
(b)
Raman intensity [a.u.]
600
600
580
Y Data
400
400
200
200
2500
3000
2500
3000
X Data
Raman shift [cm-1]
1
1
1
1
1
1
vs
vs
vs
vs
vs
vs
600
Col
Col
Col
Col
Col
Col
2
3
4
5
6
7
3500
3500
4000
4000
2D Graph 2
Wavelength [nm]
620
640
660
Solution
A
B
C
D
E
00
D2O filter
-200
-200
1500
1500
2000
45000
40000
40000
2500
2000
(c)
100
Col
Col
Col
Col
Col
Col
2500
1
1
1
1
1
1
vs Col
vs Col
vs Col
vs Col
2
vs Col
vs Col
H2O filter
3000
X Data
3000
3500
3500
4000
4000
Raman shift [cm-1]
2
3
4
5
6
7
2D Graph 2
H O concentration [mol%]
75
50
25
0
25
0.50
50
0.75
75
1.00
100
H2O filter
D2O filter
35000
30000
30000
Y Data
Integrated Raman intensity [a.u.]
620
Solution
A
B
C
D
E
2500
Y Data
600
25000
20000
20000
15000
10000
10000
0
0.00
0.25
D2O concentration
[mol%]
X Data
1.25
Col 1 vs Col 2
Fig. 6 (a) Total internal reflection Raman
spectra from microchannel filled with H2O/D2O mixtures and
Col 1 vs Col 3
x column
vs y column 1calculated by subtracting the TIR spectrum obtained
ambient air (b) Difference Raman spectra
of1 mixtures
x column 3 vs y column 3
with an empty channel (c) The relationship between the concentration and the integrated Raman intensity in
the transmittance ranges of the selected filter
-7-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
4. Calibration Experiment
4. 1 Experimental Condition
Utilizing the selected filters, a calibration experiment was carried out under a uniform concentration field
in a microchannel. The channel was positioned on the prisms along the direction of laser propagation (as
shown in Figure 5). The laser power was 0.43 W at the sample stage with an ellipsoidal spot (major axis: 260
µm, minor axis: 100 µm). The exposure time of the camera was set to be 3 s and 10 images were
accumulated for each condition.
4. 2 Calibration Result
Figure 7 shows the temporally averaged Raman images of each mixture obtained through the H2O and
D2O filters. These intensity distributions are visualized at a spatial resolution of 1.6 × 1.6 µm2 which was
calculated based on the pixel size of the camera and the total magnification of the system. It is clearly seen
that the intensity detected through the H2O filter decreases with a decreases in H2O concentration (Figure
7(a)), whereas that obtained through the D2O filter increased (Figure 7(b)). This result indicates that these
two filters are adequate for concentration imaging of the corresponding species. Although these images were
acquired under uniform concentration, they have non-uniform intensity distributions according to the
ellipsoidal intensity profile of the evanescent wave spot.
In order to obtain the relationship between the Raman intensity and the concentration of each species,
the intensities at arbitrary ten points in the Raman images, which were randomly selected from the region of
evanescent wave spot, are plotted against the concentration (in Figure 8). For reducing the influence of the
inhomogeneous intensity profile, as indicated in Eq (4), the Raman intensities were normalized by the
reference intensity being the one of at 50 mol%. The normalized intensities obtained for the H2O and D2O
filters linearly increase with H2O and D2O concentrations, respectively. Linear least-squares fit provides the
solid straight lines in Figure 8, which served as calibration curves for subsequent concentration
measurement. The error bars represent the standard deviation of intensity ratios between the ten points. By
applying this calibration result, Raman images can be converted into concentration maps of H2O and D2O.
(a) H2O filter
Solution A
Solution C
Solution E
Intensity [a.u.]
× 104
1.0
0.8
0.6
0.4
0.2
130 µm
(b) D2O filter
Evanescent wave spot
Solution A
Solution C
Solution E
Intensity [a.u.]
× 104
1.4
1.2
1.0
0.8
0.6
0.4
0.2
Fig. 7 Raman images of each mixture obtained through (a) H2O filter and (b) D2O filter
-8-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
1.8
1.8
100
75
50
0
25
H2O filter
D2O filter
1.6
1.6
1.4
1.4
1.2
1.2
Y Data
Normalized Raman intensity [-]
2D Graph 1 [mol%]
H2O concentration
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
00
25
25
50
50
75
75
100
100
125
X Data
D2O concentration
[mol%]
1 vs Col 2
Fig. 8 Calibration curves Col
between
the normalized Raman intensities and the concentration
Col 1 vs Col 4
x column 2 vs y column 2
x column 3 vs y column 3
5. Near-wall Concentration Imaging in Mixing Flow
Using the above calibration result, near-wall concentration measurement was performed in the nonuniform mixing flow field in a T-shaped microchannel (Figure 5). Pure H2O and D2O were injected into the
channel by pressure-driven flow from Inlets A and B, respectively. The position of the second evanescent
wave spot was adjusted to the junction area where the two liquids merged (Figure 9(a)). The conditions for
the data acquisition (such as laser poser, exposure time and the frame number of accumulation) were set to
be the same as the calibration experiment. Figures 9(b) and (c) give the measurement results for H2O and
D2O concentrations, respectively. It should be noted that the visualized concentration maps represent the
‘initial concentration’ without consideration of isotopic exchange reaction (H2O + D2O ↔ 2HOD). The nonuniform concentration distributions were visualized for each species in the evanescent wave spot using the
corresponding filter and calibration curve. The concentration gradient can be successfully observed in the Xdirection in the central region of both distributions although the concentration around the edge of the spot is
considered to suffer from lower signal level. These results clearly demonstrate the availability of TIR Raman
imaging for visualization of molecular distributions in the vicinity of the channel wall.
(a)
(b) H2O concentration
Evanescent wave spot
Major axis: 260 µm
Minor axis: 100 µm
(c) D2O concentration
D2O
H2O
Y
Light path
X
[mol%]
100
80
60
40
20
0
Fig. 9 (a) Measurement area at the junction of the T-shaped microchannel. (b) Two-dimensional distribution of H2O
concentration and (c) of D2O concentration.
-9-
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
6. Conclusions
The present study describes the development of a non-intrusive visualization technique for near-wall
concentration distribution using total internal reflection Raman imaging. The evanescent wave, which was
generated by a two-prism-based illumination system, was used as the excitation light source for Raman
process at solid-liquid interface in a microchannel. H2O/D2O mixtures were prepared as sample solutions and
optical bandpass filters were selected based on their TIR Raman spectra for concentration imaging. The main
accomplishments obtained from this work are summarized below.
(1) A calibration experiment was carried out under a uniform concentration field. The Raman scattering
intensity obtained through the selected filters linearly increased with the concentration of the
corresponding species.
(2) A near-wall concentration distribution measurement was performed in the mixing flow field in a Tshaped microchannel. The non-uniform concentration map in the evanescent wave spot was visualized
for each species at a spatial resolution of 1.6 × 1.6 µm2 with the penetration depth of around 90 nm.
It can be concluded that the results obtained in this study prove the feasibility of the TIR Raman imaging
technique for quantitative visualization of near-wall molecular distributions for analyzing surface-related
phenomena.
Acknowledgements
This work was supported by a Grant-in-Aid for Scientific Research (A) (No. 23246037), a Grant-in-Aid
for JSPS Fellows (No. 24-7201) (for the first author) from the Japan Society for the Promotion of Science
and Keio Gijuku Academic Development Funds.
References
Arora A, Simone G, Salieb-Beugelaar GB, Kim JT, Manz A (2010) Latest Development in Micro Total
Analysis Systems. Anal Chem 82: 4830–4847.
Axelrod D (2001) Total Internal Reflection Fluorescence Microscopy in Cell Biology. Traffic 2:764–774.
Hu Q, Tay LL, Noestheden M, Pezacki P (2007) Mammalian Cell Surface Imaging with NitrileFunctionalized Nanoprobes   Biophysical Characterization of Aggregation and Polarization Anisotropy in
SERS Imaging. J Am Chem Soc 129: 14–15.
Ikeshoji T, Ono Y, Mizuno T (1973) Total Reflection Raman Spectra; Raman Scattering due to the
Evanescent Wave in Total Reflection. Appl Opt 12: 2236–2237.
Kazoe Y, Sato Y (2007) Effect of Ion Motion on Zeta-Potential Distribution at Microchannel Wall Obtained
from Nanoscale Laser-Induced Fluorescence. Anal Chem 79:6727–6733.
Kazoe Y, Miyakawa S, Miki N, Sato Y (2009) Fluorescence Imaging Technique of Surface Electrostatic
Potential using Evanescent Wave Illumination. Appl Phys Lett 95: 234104.
Kazoe Y, Nakamura T, Sato Y (2010) Evanescent-wave / Volume Illuminated Velocity Measurements of
Transient Electrokinetically Driven Flow with Nonuniform Wall Electrostatic Potential. Meas Sci Technol
21: 055401.
Kim H, Kosuda KM, Duyne RPV, Stair PC (2010) Resonance Raman and Surface- and Tip-enhanced
Raman Spectroscopy Methods to Study Solid Catalysts and Heterogeneous Catalytic Reactions. Chem Soc
Rev 39: 4820–4844.
- 10 -
th
17 International Symposium on Applications of Laser Techniques to Fluid Mechanics
Lisbon, Portugal, 07-10 July, 2014
Kudelski A (2009) Raman Spectroscopy of Surfaces. Surf Sci 603: 1328–1334.
Nickolov ZS, Earnshaw JC, McGarvey JJ (1993) Water Structure at Interfaces Studied by Total Internal
Reflection Raman Spectroscopy. Colloids Surfaces A 76: 41–49.
Piorek BD, Lee SJ, Santiago JG, Moskovits M, Banerjee S, Meinhart CD (2007) Free-surface Microfluidic
Control of Surface-enhanced Raman Spectroscopy for the Optimized Detection of Airborne Molecules. Proc
Natl Acad Sci USA 104: 18898–18901.
Probstein RF (1994) Physicochemical Hydrodynamics, An Introduction second edition. John Willy & Sons.
Rinke G, Wenka A, Roetmann K, Wackerbarth H (2012) In Situ Raman Imaging Combined with
Computational Fluid Dynamics for Measuring Concentration Profiles during Mixing Processes. Chem Eng J
179: 338–348.
Sadr R, Yoda M, Zheng Z, Conlisk AT (2004) An Experimental Study of Electro-osmotic Flow in
Rectangular Microchannels. J Fluid Mech 506: 357–367.
Savitzky A, Golay MJE (1964) Smoothing and Differentiation of Data by Simplified Least Squares
Procedures. Anal Chem 36: 1627–1639.
Takahashi M, Furukawa T, Sato Y, Hishida K (2012) Non-Intrusive Velocity Measurement of Millichannel
Flow by Spontaneous Raman Imaging. J Therm Sci Technol 7: 406–413.
Tyrode E, Rutland MW, Bain CD (2008) Adsorption of CTAB on Hydrophilic Silica Studied by Linear and
Nonlinear Optical Spectroscopy. J Am Chem Soc 130: 17434–17445.
Woods DA, Bain CD (2012) Total Internal Reflection Raman Spectroscopy. Analyst 137:35–48.
Xia Y, Whitesides GM (1998) Soft Lithography. Annu Rev Mater Sci 28: 153–184.
Zettner CM, Yoda M (2003) Particle Velocity Field Measurements in a Near-wall Flow using Evanescent
Wave Illumination. Exp Fluids 34: 115–121.
- 11 -