sensors
Article
A Spatially Offset Raman Spectroscopy Method
for Non-Destructive Detection of
Gelatin-Encapsulated Powders
Kuanglin Chao 1, *, Sagar Dhakal 1 , Jianwei Qin 1 , Yankun Peng 2 , Walter F. Schmidt 1 ,
Moon S. Kim 1 and Diane E. Chan 1
1
2
*
Environmental Microbial and Food Safety Laboratory, Agricultural Research Service, United States
Department of Agriculture, 10300 Baltimore Avenue, Building 303 BARC-East, Beltsville, MD 20705, USA;
[email protected] (S.D); [email protected] (J.Q.); [email protected] (W.F.S.);
[email protected] (M.S.K.); [email protected] (D.E.C.)
National R&D Centre for Agro-Processing, China Agriculture University, 17 Qinghua East Road, Haidian,
Beijing 100083, China; [email protected]
Correspondence: [email protected]; Tel.: +1-301-504-8450; Fax: +1-301-504-9466
Academic Editors: Sebastian Wachsmann-Hogiu and Zachary J. Smith
Received: 26 January 2017; Accepted: 16 March 2017; Published: 18 March 2017
Abstract: Non-destructive subsurface detection of encapsulated, coated, or seal-packaged foods and
pharmaceuticals can help prevent distribution and consumption of counterfeit or hazardous products.
This study used a Spatially Offset Raman Spectroscopy (SORS) method to detect and identify urea,
ibuprofen, and acetaminophen powders contained within one or more (up to eight) layers of gelatin
capsules to demonstrate subsurface chemical detection and identification. A 785-nm point-scan
Raman spectroscopy system was used to acquire spatially offset Raman spectra for an offset range
of 0 to 10 mm from the surfaces of 24 encapsulated samples, using a step size of 0.1 mm to obtain
101 spectral measurements per sample. As the offset distance was increased, the spectral contribution
from the subsurface powder gradually outweighed that of the surface capsule layers, allowing for
detection of the encapsulated powders. Containing mixed contributions from the powder and capsule,
the SORS spectra for each sample were resolved into pure component spectra using self-modeling
mixture analysis (SMA) and the corresponding components were identified using spectral information
divergence values. As demonstrated here for detecting chemicals contained inside thick capsule
layers, this SORS measurement technique coupled with SMA has the potential to be a reliable
non-destructive method for subsurface inspection and authentication of foods, health supplements,
and pharmaceutical products that are prepared or packaged with semi-transparent materials.
Keywords: spatially offset Raman spectroscopy; self-modeling mixture analysis; subsurface detection;
quality control
1. Introduction
Raman spectroscopy has been demonstrated to be a reliable sensing technique for identification
and authentication of many materials [1]. In particular, the higher sensitivity and chemical specificity
of Raman spectroscopy are strong advantages over other analytical methods [2]. Raman spectroscopy
technique has witnessed rapid advancement in recent years, and the growing interests of research
and industry have further boosted its application. In recent years, rapid development and growing
interest from both industry and research have spread the use of Raman spectroscopy to a wide variety
of new applications, including areas in food safety and quality detection [3,4], pharmaceutical quality
control [5], and forensic and biomedical analyses [6,7].
Sensors 2017, 17, 618; doi:10.3390/s17030618
www.mdpi.com/journal/sensors
Sensors 2017, 17, 618
2 of 12
Conventional backscattering Raman spectroscopy technique is suitable for surface analysis, but,
in many cases, its use for analyzing a subsurface material through another surface layer is commonly
ineffective or difficult due to an overwhelming Raman and/or fluorescence signal from the surface
layer material [8]. Materials of heterogenous composition can also present difficulties due to the
limited area and depth that can be analyzed. Transmission Raman spectroscopy can overcome
both subsurface presentation and bulk heterogeneity issues, allowing for retrieval of some Raman
information from within a sample, by placing the sample between a laser and detector in order
to acquire forward-scattered Raman signals that have passed through the sample. Transmission
Raman spectroscopy has been effectively used for internal analysis of samples such as quantitative
assessment of pharmaceutical capsules [9,10], evaluation of protein content in packed corn kernels [11],
and evaluation of protein and oil composition in single soybeans [12]. Although this technique can
overcome surface layer fluorescence (e.g., of a capsule, coating, or other packaging layer, etc.) for
analysis of internal layers, the mixed Raman information retrieved from multiple internal layers cannot
be separated.
Spatially offset Raman spectroscopy (SORS) is a technique that can retrieve subsurface Raman
information from diffusely scattering media [13]. Raman signals are collected along the sample surface
at a series of points spatially offset from the point of illumination. Increasing the spatial offset increases
the signal contribution from deeper layers such that they gradually outweigh the signal from the top
surface material, thereby enhancing the internal signal while attenuating the surface signal. The pattern
of spectral change that occurs with increasing offset distance allows for retrieval of Raman signals from
multiple different internal layers within the sample. A preliminary study investigated the potential
of the SORS technique for detection of concealed drugs [14]. SORS technique is also applied for
biomedical analysis [15,16], safety and quality analysis of food materials [17–19], and forensics [20].
Incidents of contamination and adulteration in food and pharmaceutical products—such as
capsule tampering [21], adulteration of dietary supplement capsules [22], and adulteration of food
powders [23]—has necessitated the development of non-invasive methods to detect and identify
components within sealed containers. SORS has already been demonstrated for retrieving information
from deep layered samples, but, used alone, is not sufficient to characterize the components at different
individual layers. SORS data contains mixed spectral information from materials in the upper surface
layer and in the deeper subsurface layers. This mixed spectral information must be resolved into
separate pure component spectra to identify individual components at each layer. One method to
resolve the mixed spectral data matrix is self-modeling mixture analysis (SMA), a serial algorithm
that uses an alternating least squares approach to decompose the mixed spectral data to obtain pure
component spectra and their corresponding contributions [24,25]. SMA has been effectively used to
retrieve pure component spectra from mixed spectral data for sample mixtures such as melamine, urea,
ammonium sulfate, and dicyandiamide mixed with dry milk [26]; and vanillin, sugar and melamine
mixed with non-dairy powdered creamer [27].
This study demonstrates a point-scan SORS method to retrieve Raman spectral information
for three chemicals (urea, ibuprofen, and acetaminophen) encapsulated within multiple layers of
gelatin capsules. By using a motorized positioning table to precisely control offset distances for SORS
collection, the offset distance and offset interval can be easily customized for different applications.
The SMA algorithm was applied to characterize each component in the capsule-chemical samples.
The presented method has potential application for food safety and quality detection, pharmaceutical
quality control, and forensic sciences. The main objectives of this study were to:
1.
2.
3.
Develop a method to acquire spatially offset Raman spectra using a point-scan Raman
spectroscopy system;
Create capsule-chemical samples comprised of up to eight gelatin capsule layers, and acquire
and analyze SORS data with respect to capsule layers and offset distances;
Use self-modeling mixture analysis to extract pure component spectra from each capsule-chemical
sample and identify chemical components corresponding to the extracted spectra.
2. Materials and Methods
2.1. Spatially
Offset Raman Spectroscopy System
Sensors 2017, 17, 618
3 of 12
SORS sample measurements were collected in the wavenumber range of 107 to 2560 cm−1 using
2. Materials
andspectroscopy
Methods
a point-scan
Raman
system set up as follows. A charge-coupled device (CCD) camera
with 16 bit depth and 1024 × 256 active pixels (Newton DU920N-BR-DD, Andor Technology, South
2.1. Spatially Offset Raman Spectroscopy System
Windor, CT, USA) was mounted with a Raman spectrometer (Raman Explorer 785, Headwall
SORS sample measurements were collected in the wavenumber range of 107 to 2560 cm−1 using
Photonics, Fitchburg, MA, USA) for signal acquisition. For sample excitation, light from a 785-nm
a point-scan Raman spectroscopy system set up as follows. A charge-coupled device (CCD) camera
laser module
Innovative
Photonics
Solutions, Monmouth
Junction, South
NJ, USA) was
with 16(I0785MM0500MF,
bit depth and 1024 × 256
active pixels
(Newton DU920N-BR-DD,
Andor Technology,
delivered
by optical
fiber
focused
thespectrometer
sample surface
a fixed785,
45°
incident
angle by a laser
Windor,
CT, USA)
wasand
mounted
with aonto
Raman
(Ramanat
Explorer
Headwall
Photonics,
Fitchburg,
MA,
for signal
acquisition. Forofsample
excitation,
from
a 785-nm
laser module
focus unit.
Figure
1aUSA)
shows
the arrangement
the laser
focuslight
unit,
sample
platform,
and Raman
(I0785MM0500MF,
Innovative
PhotonicsFigure
Solutions,
USA) was delivered
probe within
the sample
compartment.
1b Monmouth
shows theJunction,
opticalNJ,
components
of the by
laser focus
optical fiber and focused onto the sample surface at a fixed 45◦ incident angle by a laser focus unit.
unit; first collimated by the fiber optic collimator, the laser light is filtered of erroneous background
Figure 1a shows the arrangement of the laser focus unit, sample platform, and Raman probe within the
light and
purified
by the 785
nm1blaser-line
filter and
then
focused
onto
sample surface
sample
compartment.
Figure
shows the bandpass
optical components
of the
laser
focus unit;
firstthe
collimated
by the focus
probe
(RPB,
Norwood,
USA) was fixed
by thelens.
fiber The
optic fiber-optical
collimator, theRaman
laser light
is filtered
of InPhotonics,
erroneous background
lightMA,
and purified
by
the
785
nm
laser-line
bandpass
filter
and
then
focused
onto
the
sample
surface
by
the
focus
lens.
perpendicular to the sample surface to collect and transfer Raman signals to the spectrometer
(RPB,
Norwood,
MA, USA) was
perpendicular
to used as
throughThe
anfiber-optical
input slit Raman
(5 mmprobe
length
× InPhotonics,
100 µ m width).
Polystyrene
andfixed
naphthalene
were
the sample surface to collect and transfer Raman signals to the spectrometer through an input slit −1
Raman shift standards for spectral calibration of the system. Spectral resolution was 14 cm at full
(5 mm length × 100 µm width). Polystyrene and naphthalene were used as Raman shift standards for
width half
maximum.
spectral
calibration of the system. Spectral resolution was 14 cm−1 at full width half maximum.
To input slit of
Raman spectrometer
Optical
fiber
To laser source
Laser focus unit
Optical probe
(fixed position)
Stand
Sample
Motorized
platform
Platform moving direction
(a)
Laser focus unit
Adaptor
Raman scattering signal
FC/PC
connector
Optical
fiber
Focus
lens
Collimator
lens
Laser line
filter
(b)
Incident laser
Capsule surface
Chemical layer
x
Offset distance
(c)
Figure Figure
1. (a) 1.Sample
compartment
SORS
data collection;
(b) unit;
laserandfocus
unit;
and (c) SORS
(a) Sample
compartment forfor
SORS
data collection;
(b) laser focus
(c) SORS
technique
for detecting
RamanRaman
scattering
from sample
depths.
technique
for detecting
scattering
from
sample depths.
Bothsample
the sample
laserfocus
focus units
units were
onto
a motorized
positioning
platform platform
Both the
andand
laser
werefixed
fixed
onto
a motorized
positioning
(MAXY4009W1-S4, Velmex, Bloomfield, NY, USA) for synchronized movement, to maintain the fixed
(MAXY4009W1-S4,
Velmex, Bloomfield, NY, USA) for synchronized movement, to maintain the
excitation spot on the sample surface for collection of Raman spectral signal at different offset distances
fixed excitation
spot on the sample surface for collection of Raman spectral signal at different offset
using the fixed-position Raman probe. Camera control, spectral data acquisition, and movement of
distances
theplatform
fixed-position
Raman
data (National
acquisition, and
the using
motorized
were performed
by probe.
softwareCamera
developedcontrol,
in-house spectral
using LabView
movement
of theAustin,
motorized
platform
were
performed
software
in-house
using
Instruments,
TX, USA).
The system
was set
up to acquireby
an initial
Ramandeveloped
spectrum at no
offset,
LabView (National Instruments, Austin, TX, USA). The system was set up to acquire an initial
Raman spectrum at no offset, and then move the sample and laser focus unit (in the direction shown in
Sensors 2017, 17, 618
4 of 12
and then move the sample and laser focus unit (in the direction shown in Figure 1a) to incrementally
increase the offset distance at which each subsequent spectrum in the series is measured, thereby
obtaining a set of spatially offset Raman spectra for a single sample using the fixed-position Raman
probe. Figure 1c shows how the SORS technique can acquire Raman scattering from increasing sample
depths as the offset distance is increased. To avoid the influence of ambient light, the entire sample
compartment was enclosed within a black box.
2.2. Experimental Samples and SORS Data Acquisition
Three chemicals in powder form—urea (>98%), ibuprofen (>98%), and acetaminophen
(>99%)—were obtained from Sigma-Aldrich (St. Louis, MO, USA), and empty gelatin capsules
(single-wall thickness 0.11 mm) were obtained from Capsuline (Pompano Beach, FL, USA), to prepare
eight capsule-chemical samples for each of the three chemical powders. Gelatin capsules of size 000
(length 26.1 mm), size 00 (length 23.4 mm), and size 0 (length 21.6 mm) were used to prepare capsule
layers. A 1-layer capsule was prepared by packing a single 000 capsule tightly with powder and
capping it. A 2-layer capsule was prepared by nesting an uncapped capsule inside another, packing the
innermost capsule with powder, and capping only the outer capsule. In similar fashion, three-, four-,
five-, six-, seven-, and eight-layer capsules were prepared by nesting uncapped capsules, packing
the deepest capsule with powder, and capping the outermost capsule. A total of 24 capsule-chemical
samples, with capsule wall thicknesses ranging from 0.11 mm (one-layer capsule) to 0.88 mm
(eight-layer capsule), were prepared for this study.
Individual Raman spectra of an empty gelatin capsule and of unobscured samples of urea,
ibuprofen, and acetaminophen were acquired as reference measurements. To avoid sample movement
during SORS measurement, each capsule-chemical sample was held immobile within a nickel-plated
sample holder and was affixed to the positioning platform. Laser light (250 mW) was focused at a spot
on the capsule body near the edge of the capsule cap. After acquiring the first Raman spectrum with no
offset, the platform moved the sample and laser focus unit in incremental steps away from the probe
such that the probe was no longer directly above the laser focus point on the sample, thus creating an
increased offset distance for each subsequent spectral acquisition. A series of SORS measurements
along the longitudinal flat surface of each capsule-chemical sample were collected in single scan
mode with an exposure time of 0.2 s, using step size of 0.1 mm for the offset range of 0 (no offset) to
10 mm, proceeding away from the capped end of the sample. A total of 101 spectra were collected for
each sample.
2.3. SORS Data Analysis
Spectral analysis was performed using Matlab (MathWorks, Natick, MA, USA). To reduce
computational complexity, only data in the range of 500 cm−1 to 2000 cm−1 were analyzed since
this range carried significant information for the capsules and chemicals used in this study. Since the
laser-sample interactions generated fluorescence signals during SORS measurements and the higher
intensity fluorescence background can overwhelm and diminish the weak Raman peaks of interest in
the spectra, all spectra were background-corrected using a polynomial curve-fitting method [23] prior
to analysis.
Containing mixed spectral contributions from the surface capsule layers and the subsurface
chemical powders, the background-corrected SORS spectra for each capsule-chemical sample were
analyzed to examine the contributions of each that could be detected at different offset distances.
The SORS dataset for each capsule-chemical sample was decomposed using self-modeling mixture
analysis (SMA) to obtain pure component spectra of capsule and chemicals. The key to using SMA
is the presence of a pure variable in the mixed spectral data; for a SORS dataset, this is a Raman
wavenumber at which only one component contributes significant signal intensity. SMA involves one
step to determine pure variables and a second step to resolve mixed spectral data into pure component
spectra (and contributions). The first step requires obtaining a series of purity spectra, which begins
Sensors 2017, 17, 618
5 of 12
with calculation
Sensors
2017, 17, 618 of the average and standard deviation spectra for the data being analyzed; dividing
5 ofthe
12
standard deviation spectrum by the average spectrum produces the first purity spectrum. The variable
found
towavenumber)
be the first pure
Next, intensity
a determinant-based
weight
function
is the
obtained
by
(Raman
with variable.
the maximum
in this spectrum
is found
to be
first pure
calculating
the correlation
matrix from
the mixed
spectral
data matrix.
The first
spectrum
is
variable. Next,
a determinant-based
weight
function
is obtained
by calculating
thepurity
correlation
matrix
multiplied
by the
weight
to obtain
the spectrum
second purity
spectrum.
Again,
Raman
from the mixed
spectral
datafunction
matrix. The
first purity
is multiplied
by the
weightthe
function
to
wavenumber
with purity
the maximum
intensity
theRaman
secondwavenumber
purity spectrum
then
found as intensity
the second
obtain the second
spectrum.
Again,inthe
withisthe
maximum
in
pure
variable.
Thespectrum
process is
is then
repeated
purity
spectra
no longer
exhibit
spectral features
the second
purity
founduntil
as thethe
second
pure
variable.
The process
is repeated
until the
[24,25,28].
A correction
offset,
is added
to the A
average
spectrum
reduce
the is
effect
of
purity spectra
no longerfactor,
exhibitcalled
spectral
features
[24,25,28].
correction
factor,tocalled
offset,
added
noise
calculating
thetopurity
After
the pure
variables
have the
beenpurity
determined,
mixed
to thewhen
average
spectrum
reducespectra.
the effect
of noise
when
calculating
spectra.the
After
the
spectra
datasethave
can be
resolved
into pure
component
corresponding
by
pure variables
been
determined,
the mixed
spectraspectra
datasetand
can be
resolved intocontributions
pure component
alternating
least squares contributions
method [24,25].
The 101 least
× 975
mixed
spectral
matrix
for ×each
spectra and corresponding
by alternating
squares
method
[24,25].
The 101
975
capsule-chemical
sample
wascapsule-chemical
decomposed using
function
PLS_Toolbox
mixed spectral matrix
for each
samplethe
was purity
decomposed
usinginthethe
purity
function in
(Eigenvector
Research,
Inc., Wenatchee,
USA) to obtain
spectra
of capsule
and
the PLS_Toolbox
(Eigenvector
Research, WA,
Inc., Wenatchee,
WA,pure
USA)component
to obtain pure
component
spectra
chemicals.
of capsule and chemicals.
After
After SMA,
SMA, spectral
spectral information
information divergence
divergence (SID)
(SID) was
was used
used to
to match
match pure
pure component
component spectra
spectra
to
to their
their corresponding
corresponding components
components by
by comparing
comparing the
the dissimilarity
dissimilarity between
between spectra
spectra by
by relative
relative
entropy
× 1)
entropy [29].
[29]. Probability
Probability vectors
vectors obtained
obtained for
for each
each reference spectrum (975 ×
1) and
and each
each spectrum
spectrum
resolved
resolved from
from SMA
SMA (975
(975 ××1)1)were
wereused
usedtotodetermine
determinethe
therelative
relativeentropy
entropy(i.e.,
(i.e.,information
informationdivergence)
divergence)
of
of the
the reference
reference spectrum
spectrum with
with respect
respect to
to SMA-resolved
SMA-resolved spectrum,
spectrum, and
and vice
vice versa.
versa. Summation
Summation of
of the
the
two
two relative
relative entropies
entropies gives
gives the
the SID
SID value.
value. A
A smaller
smaller SID
SID value
value indicates
indicates less
less disparity
disparity between
between the
the
spectra
thus aabetter
bettermatch.
match.Based
Based
values,
extracted
spectrum
was identified
as
spectra and thus
onon
SIDSID
values,
eacheach
extracted
spectrum
was identified
as being
being
urea, ibuprofen,
acetaminophen,
or Identification
capsule. Identification
of each corresponding
that ofthat
urea,ofibuprofen,
acetaminophen,
or capsule.
of each corresponding
component
component
by SIDtoalso
serves
validate
the of
SMA
method
of retrieving
component
spectra
by SID also serves
validate
thetoSMA
method
retrieving
pure
componentpure
spectra
from the original
from
the
original mixed spectra.
mixed
spectra.
3. Results
Resultsand
andDiscussions
Discussions
3.
3.1. Spectral
Spectral and
and Spatial
Spatial Chracteristics
Chracteristics of
of Samples
Samples
3.1.
Figure 22 shows
shows the
the chemical
chemical structures
structures of
of urea,
urea, ibuprofen
ibuprofen and
and acetaminophen.
acetaminophen. Figure
Figure 33 shows
shows
Figure
the
reference
spectra
of
urea,
ibuprofen,
acetaminophen,
and
gelatin
capsule.
Urea,
chemical
formula
the reference spectra of urea, ibuprofen, acetaminophen, and gelatin capsule. Urea, chemical
CO(NH2CO(NH
)2 , has two
–NH2 groups
a carbonyl
(C=O) functional
group, asgroup,
shownas
in shown
Figure in
2a.
formula
2)2, has two
–NH2 joined
groupsbyjoined
by a carbonyl
(C=O) functional
−1 due to (C–N–C) stretching is most definitive to its
The
high
intensity
peak
of
urea
at
1009
cm
Figure 2a. The high intensity peak of urea at 1009 cm−1 due to (C–N–C) stretching is most definitive
−1
identification
and and
quantification
[30,31],
while
its other
peaks
(not
labeled
ininFigure
to
its identification
quantification
[30,31],
while
its other
peaks
(not
labeled
Figure3:3:1646
1646cm
cm−1
−
1
−
1
from asymmetrical
asymmetrical bending,
bending, 1541
from NH
NH symmetrical
symmetrical bending,
bending, and
and 1175
from H–N–H
H–N–H
from
1541 cm
cm−1 from
1175 cm
cm−1 from
bending)
are
much
less
prominent.
Ibuprofen
(Figure
2b)
has
carboxylic
acid
and
an
aromatic
ring as
as
bending) are much less prominent. Ibuprofen (Figure 2b) has carboxylic acid and an aromatic ring
functional groups;
groups; acetaminophen
acetaminophen(Figure
(Figure 2c)
2c) also
also has
has an
an aromatic
aromatic ring.
ring. Both
Both the
the ibuprofen
ibuprofen peak
peak at
at
functional
−1 and the vibrational peak of acetaminophen at 859 cm−1 are due to ring deformation [32–34].
833
cm
−1
−1
833 cm and the vibrational peak of acetaminophen at 859 cm are due to ring deformation [32–34].
−1 is due to tyrosine [35,36].
The gelatin
gelatin capsule
capsule peak
peak at
at 644
644 cm
cm−1
is due to tyrosine [35,36].
The
(a)
(b)
(c)
Figure 2. Chemical structures of (a) urea; (b) ibuprofen; and (c) acetaminophen. Oxygen is
Figure 2. Chemical structures of (a) urea; (b) ibuprofen; and (c) acetaminophen. Oxygen is highlighted
highlighted in red, nitrogen in dark blue, carbon in light blue, and hydrogen in white.
in red, nitrogen in dark blue, carbon in light blue, and hydrogen in white.
Sensors 2017, 17, 618
6 of 12
Sensors 2017, 17, 618
6 of 12
Sensors 2017, 17, 618
6 of 12
Figure
Figure 3.
3. Reference
Reference Raman
Raman spectra
spectra of
of urea,
urea, ibuprofen,
ibuprofen, acetaminophen,
acetaminophen, and
and gelatin
gelatin capsule.
capsule.
3. Reference
urea, ibuprofen,
acetaminophen,
gelatin capsule.
FigureFigure
4a shows
a 1024Raman
× 256spectra
pixel of
Raman
scattering
image for aand
one-layer
capsule of urea,
Figure 4a shows a 1024 × 256 pixel
Raman
scattering
image
for
a
one-layer
capsule of urea,
−1
−1
covering a Raman shift range of 250 cm −1to 1885 cm −and
spatial range of 10 mm (corresponding to
1
Figure
4a
shows
a
1024
×
256
pixel
Raman
scattering
image
for
a
one-layer
capsule
of urea,
covering a Raman shift range of 250 cm to 1885 cm and spatial range of 10 mm (corresponding
the spatial offset range from 0 to 10 mm −1on the capsule-chemical
sample surface. The top 0 to 3 mm
−1
a Raman
range
cm mm
to 1885
cm capsule-chemical
and spatial rangesample
of 10 mm
(corresponding
to covering
the spatial
offset shift
range
fromof0250
to 10
on the
surface.
The top to
0 to
on the Y-axis is presented merely for clarity of the image. The scattering image represents spectral
the spatial
range
from 0 to 10
mm on
capsule-chemical
sample
surface. Theimage
top 0 to
3 mm
3 mm
on theoffset
Y-axis
is presented
merely
forthe
clarity
of the image.
The scattering
represents
and
spatial
information
of the
SORSfor
measurement
of capsule with
urea. Any
vertical
line parallel
to
on theand
Y-axis
is presented
merely
the image. The
image
represents
spectral
spectral
spatial
information
of theclarity
SORSof
measurement
of scattering
capsule with
urea.
Any vertical
line
theand
spatial
axis
represents
a
spatial
profile
at
the
corresponding
wavenumber.
For
example,
the
two
the SORSameasurement
with urea. Any
vertical line For
parallel
to
parallelspatial
to theinformation
spatial axis of
represents
spatial profileofatcapsule
the corresponding
wavenumber.
example,
−1 and 1009 cm−1.
vertical
lines axis
drawn
in Figure
4a represent
the
spatial
profile of the
sample at 644
cm
the
spatial
represents
a
spatial
profile
at
the
corresponding
wavenumber.
For
example,
the
two
the two vertical lines drawn in Figure 4a represent the spatial profile of the sample
at 644 cm−1 and
−1 and 1009 cm−1. The
Figure
4b−1lines
shows
the original,
spatial
profiles
at 644
−1 and 1009 cm−1.
vertical
drawn
in Figureuncorrected
4a represent the
spatial
profileextracted
of the sample
at cm
644 cm
1009 cm . Figure
4b shows the original, uncorrected
spatial profiles extracted at 644 cm−1 and
−1 represents urea. Because
profile
at−4b
644shows
cm−1 represents
and 1009
cmprofiles
Figure
the original,capsule
uncorrected
spatial
extracted at 644 cm−1 this
and scattering
1009 cm−1. image
The
1
−
1
1009 cm . The profile
at 644 cm represents capsule
and 1009 cm−1 represents urea. Because this
−1
−1
was
obtained
fluorescence
correction,
intensity can
observed
the baseimage
of the
profile
at 644prior
cm to
represents
capsule
and 1009 high
cm represents
urea.beBecause
this at
scattering
scattering image was obtained prior to fluorescence correction, high intensity can
be observed at the
spatial
all wavenumbers.
the offset high
distance
increases,
644 cm−1atsignal
intensity
was profile
obtainedat prior
to fluorescenceAscorrection,
intensity
can bethe
observed
the base
of the of
base of the spatial profile at all wavenumbers. As the offset distance increases,
the
644 cm−1 signal
−1 signal
thespatial
gelatin
capsule
and steeply
below
the 1009thecm
intensity
of urea,
profile
at allabruptly
wavenumbers.
As thedecreases
offset distance
increases,
644
cm−1−signal
intensity
of
intensity of the gelatin capsule abruptly and steeply decreases below the 1009
cm 1 signal intensity of
−1
the gelatin capsule
and steeply decreases
below
the 1009 chemical
cm signal
intensity that
of urea,
demonstrating
that theabruptly
signal contribution
of the deeper
subsurface
outweighs
of the
urea, demonstrating
that
the signal
contribution
the deeper
subsurface
chemical outweighs
that
the signal
contribution
theofdeeper
subsurface
chemical
that ofshows
the of
topdemonstrating
surface layer.that
Changes
in Raman
signalofcontribution
at different
spatial outweighs
offset distances
thetop
topsurface
surfacelayer.
layer.Changes
ChangesininRaman
Ramansignal
signalcontribution
contribution
at
different
spatial
offset
distances
shows
different spatial offset distances shows
the potential of this method to retrieve subsurface Raman at
information.
thethe
potential
ofof
this
information.
potential
thismethod
methodtotoretrieve
retrievesubsurface
subsurface Raman
Raman information.
(a)(a)
(b)
(b)
Figure
4. 4.
(a)(a)Raman
sample; and
and (b)
(b)spatial
spatialprofile
profileofofcapsule
capsule
Figure
Ramanscattering
scatteringimage
image of
of capsulecapsule- urea sample;
Figure −1
4. (a) Raman scattering
image of capsule- urea sample; and (b) spatial profile of capsule
−1−1
−1
(644
cmcm
)
and
urea
(1009
cm
).
(644
)
and
urea
(1009
cm
).
(644 cm−1 ) and urea (1009 cm−1 ).
Spatially
Offset
RamanSpectra
SpectraofofCapsule-Chemical
Capsule-Chemical Samples
Samples
3.2.3.2.
Spatially
Offset
Raman
3.2. Spatially Offset Raman Spectra of Capsule-Chemical Samples
Spatially
offsetRaman
Ramanspectra
spectrawere
were acquired
acquired from
from capsule-chemical
with
Spatially
offset
capsule-chemicalsamples
samplesprepared
prepared
with
Spatially
offset
Raman
spectra
were
acquired
fromcorrected
capsule-chemical
samples
prepared
with
one
to
eight
capsule
layers.
Figure
5
shows
background
spectra
for
the
one-,
three-,
five-,
one to eight capsule layers. Figure 5 shows background corrected spectra for the one-, three-, five-,
one
to
eight
capsule
layers.
Figure
5
shows
background
corrected
spectra
for
the
one-,
three-,
five-,
and
eight-layer
capsule-ureasamples
samplesto
to illustrate
illustrate the
the general
with
and
eight-layer
capsule-urea
general spectral
spectralpatterns
patternsthat
thatoccurred
occurred
with
and
eight-layer
capsule-urea
samples
to illustrate
the general
spectral
patterns
thatare
occurred
with
increasing
offset
distance
and
increasing
capsule
layers.
For
each
sample,
six
spectra
presented
increasing offset distance and increasing capsule layers. For each sample, six spectra are presented
from 0 mm offset to 10 mm offset at 2 mm intervals. All spectra were normalized and vertically
from
0 mm offset to 10 mm offset at 2 mm intervals. All spectra were normalized and vertically
Sensors 2017, 17, 618
7 of 12
increasing offset distance and increasing capsule layers. For each sample, six spectra are presented
Sensors 2017, 17, 618
7 of 12
from 0 mm offset to 10 mm offset at 2 mm intervals. All spectra were normalized and vertically shifted
forshifted
clear presentation
and comparison.
For all samples,
increasing
offset distance
attenuated
for clear presentation
and comparison.
For all
samples,the
increasing
the offset
distancethe
−1 capsule peak. For the one-layer capsule-urea sample, it can be
relative
intensity
of
the
644
cm
−1
attenuated the relative intensity of the 644 cm capsule peak. For the one-layer capsule-urea sample,
1
observed
Figure 5ainthat
the 5a
644that
cm−
is readily
at visible
0 mm offset,
attenuated
it can beinobserved
Figure
thecapsule
644 cm−1peak
capsule
peak isvisible
readily
at 0 mm
offset,
considerably
2 mm offset at
and
longer
visible apparently
after 4 mm visible
offset. Increasing
theoffset.
number
attenuated at
considerably
2 no
mm
offsetapparently
and no longer
after 4 mm
−1 capsule peak, which
−1
of Increasing
capsule layers
enhanced
the
relative
intensity
of
the
644
cm
can
be seen
the number of capsule layers enhanced the relative intensity of the 644 cm capsule peak,
bywhich
examining
peak
at the same
offset
for the
one-,
three-,for
five-,
samples
can bethis
seen
by examining
this
peakdistance
at the same
offset
distance
the and
one-,eight-layer
three-, five-,
and
eight-layer
(Figure
5a–d).samples (Figures 5a through 5d).
contrast,the
the1009
1009cm
cm−−11 urea
offset
distance.
Although
a
InIn
contrast,
urea peak
peakwas
wasenhanced
enhancedbybyincreasing
increasing
offset
distance.
Although
−1−1urea peak is observed at all offset distances for the
similar
relative
intensity
of
the
1009
cm
a similar relative intensity of the 1009 cm urea peak is observed at all offset distances for the
one-layercapsule-urea
capsule-ureasample
sample in
be be
attributed
to the
capsule
layer on
theon
one-layer
in Figure
Figure5a,
5a,this
thismay
may
attributed
to single
the single
capsule
layer
providing
minimal
suppression
of the
urea below.
For theFor
samples
prepared
with
thesurface
surface
providing
minimal
suppression
ofsubsurface
the subsurface
urea below.
the samples
prepared
multiple
capsule
layers,
gradual
enhancement
of the
urea
peak
is evident
with
increased
offset
with
multiple
capsule
layers,
gradual
enhancement
of the
urea
peak
is evident
with
increased
offset
−1 urea
distance.
The
three-layer
capsule-urea
sample
shows
a
very
small
but
identifiable
1009
cm
−
1
distance. The three-layer capsule-urea sample shows a very small but identifiable 1009 cm urea peak
peak at 0 mm offset, which is enhanced with increasing offset distance (Figure 5b). For five or more
at 0 mm offset, which is enhanced with increasing offset distance (Figure 5b). For five or more capsule
capsule layers, the 1009 cm−1 urea peak is not identifiable at 0 mm offset, as seen in Figure 5c,d,
layers, the 1009 cm−1 urea peak is not identifiable at 0 mm offset, as seen in Figure 5c,d, because the
because the deep subsurface urea signal is well suppressed by the stronger capsule signal at the
deep subsurface urea signal is well suppressed by the stronger capsule signal at the surface, and a
surface, and a no-offset measurement is equivalent to a conventional backscattering Raman
no-offset measurement is equivalent to a conventional backscattering Raman measurement. The fact
measurement. The fact that increasing the surface layer thickness prevents backscattering Raman
that
increasing the surface layer thickness prevents backscattering Raman measurement from detecting
measurement from detecting the deeper subsurface chemical shows that the conventional method is
thenot
deeper
subsurface
chemical
shows thatHowever,
the conventional
method
is not distance,
effectivethe
for1009
subsurface
effective
for subsurface
measurements.
by increasing
the offset
cm−1
−1 urea peak was gradually
measurements.
However,
by
increasing
the
offset
distance,
the
1009
cm
urea peak was gradually enhanced despite the presence of multiple layers of gelatin capsule
enhanced
the presence
of multiple
of gelatin
(Figure 5c,d),for
clearly
demonstrating
(Figure despite
5c,d), clearly
demonstrating
the layers
effectiveness
of capsule
SORS measurement
detection
of the
thedeeper
effectiveness
of SORS
measurement for detection of the deeper subsurface chemical.
subsurface
chemical.
(a)
(b)
(c)
(d)
Figure 5. Spatially offset Raman spectra for (a) one-layer; (b) three-layer; (c) five-layer; and (d)
Figure 5. Spatially offset Raman spectra for (a) one-layer; (b) three-layer; (c) five-layer; and (d)
eight-layer capsule-urea samples.
eight-layer capsule-urea samples.
Effectiveness of the SORS method was further validated with the multi-layer capsule-ibuprofen
and capsule-acetaminophen samples. The 644 cm−1 capsule peak was attenuated by increasing the
Sensors 2017, 17, 618
8 of 12
Effectiveness of the SORS method was further validated with the multi-layer capsule-ibuprofen
and capsule-acetaminophen samples. The 644 cm−1 capsule peak was attenuated by increasing
−1 ibuprofen peak and 859 cm−1 acetaminophen peak
the offset
distance
Sensors 2017,
2017,
17, 618
618 while the 833 cm
of 12
12 were
Sensors
17,
88 of
enhanced. As with the capsule-urea samples, increasing the number of capsule layers was found to
−1 ibuprofen peak and 859 cm−1
−1 acetaminophen peak were enhanced.
offsetthe
distance
while
theintensity
833 cm−1
increase
capsule
peak
when compared at the same
offset distance, and to suppress the
As with the capsule-urea samples, increasing the number of capsule layers was found to increase the
ibuprofen and acetaminophen peaks at 0 mm offset. Again, the ibuprofen and acetaminophen peak
capsule peak intensity when compared at the same offset distance, and to suppress the ibuprofen
intensities were gradually enhanced and became more prominent at larger offset distances. Figures 6
and acetaminophen peaks at 0 mm offset. Again, the ibuprofen and acetaminophen peak intensities
and 7 show spatially offset spectra measured for the one-layer and eight-layer capsule-ibuprofen
were gradually enhanced and became more prominent at larger offset distances. Figures 6 and 7
and show
capsule-acetaminophen
samples,
respectively
(for brevity,
intermediate-layer
samples
are not
spatially offset spectra
measured
for the one-layer
and eight-layer
capsule-ibuprofen
and
−1 capsule peak gradually decreased
presented).
From
Figure
6a,
it
can
be
observed
that
the
644
cm
capsule-acetaminophen samples, respectively (for brevity, intermediate-layer samples are not
−1 ibuprofen peak is similar
−1
−1
withpresented).
increasingFrom
offsetFigure
distance,
relative
intensity
833 cmpeak
6a, itwhile
can bethe
observed
that
the 644 of
cmthe
capsule
gradually decreased
−1
−1
increasing
while
relativewall
intensity
of the 833 capsule
cm ibuprofen
peak
is similar with
at allwith
offsets.
The offset
latter distance,
is due to
thinthe
capsule
at one-layer
sample.
However,
at all offsets.
TheFigure
latter is6b,
duethe
to thin
wall at one-layer
sample.
However, with the
the eight
layers in
833 capsule
cm−1 ibuprofen
peak capsule
is initially
indistinguishable
ateight
no offset,
−1
−1
layers
in
Figure
6b,
the
833
cm
ibuprofen
peak
is
initially
indistinguishable
at
no
offset,
with
theof the
with the signal of the thicker gelatin surface (approximately 0.88 mm) dominating the signal
signal
of
the
thicker
gelatin
surface
(approximately
0.88
mm)
dominating
the
signal
of
the
deeper
−
1
deeper subsurface ibuprofen. Yet, clearly, the
833 cm ibuprofen peak is gradually enhanced at
−1
subsurface ibuprofen. Yet, clearly, the 833 cm−1
ibuprofen peak is gradually enhanced at larger offset
larger offset distances for the eight-layer sample (Figure 6b). Similar patterns were observed for the
distances for the eight-layer sample (Figure 6b). Similar patterns were observed for the
capsule-acetaminophen
samples (Figure 7), all together validating SORS as an effective method for
capsule-acetaminophen samples (Figure 7), all together validating SORS as an effective method for
detecting
deeper
subsurface
chemicals.
detecting deeper subsurface
chemicals.
(a)
(b)
Figure 6. Spatially offset Raman spectra of (a) one-layer and (b) eight-layer capsule-ibuprofen
samples.
Figure 6. Spatially offset Raman spectra of (a) one-layer and (b) eight-layer capsule-ibuprofen samples.
(a)
(b)
Figure 7. Spatially offset Raman spectra of (a) one-layer and (b) eight-layer capsule-acetaminophen
samples.
Figure 7. Spatially offset Raman spectra of (a) one-layer and (b) eight-layer capsule-acetaminophen samples.
The The
weak
or nonexistent
forthe
thepowders
powders
0 mm
offset
shows
how the
weak
or nonexistentRaman
Ramanpeaks
peaks observed
observed for
at 0atmm
offset
shows
how the
surface
gelatin
layers
dominate
signaland
andattenuate
attenuate
contribution
ofsubsurface
the subsurface
surface
gelatin
layers
dominatethe
thedetectable
detectable signal
thethe
contribution
of the
powders
in a conventional
backscattering
Raman
measurement.
The
dominance
surface
powders
in a conventional
backscattering
Raman
measurement.
The
dominance
of of
thethe
surface
gelatin
gelatin
signal
was
overcome
by
separating
the
laser
focal
point
and
signal
detection
point
with
signal was overcome by separating the laser focal point and signal detection point with increasing
increasing distances, allowing the subsurface chemical contribution to gradually outweigh the
distances, allowing the subsurface chemical contribution to gradually outweigh the surface capsule
surface capsule contribution. The shift in the signal strength of the capsule component and
contribution. The shift in the signal strength of the capsule component and subsurface powder
subsurface powder component can be examined by calculating the intensity ratio between the key
Sensors 2017, 17, 618
9 of 12
component can be examined by calculating the intensity ratio between the key powder peak and
the 644Sensors
cm−12017,
capsule
A higher
17, 618 peak at each offset distance and observing the trend in ratio values.
9 of 12
ratio value indicates that the signal intensity of the subsurface powder component outweighed that of
Sensors
2017, 17,
618and the 644 cm−1 capsule peak at each offset distance and observing the trend in ratio
9 of 12
powder
peak
the capsule. Figure 8 shows the ratio of peak intensities at 1009 cm−1 (urea) and 644 cm−1 (capsule)
values. A higher ratio value indicates that the signal intensity of the subsurface powder component
as a function
of offset
0.5 mm
intervals
up of
to
3 mm)
on observing
theatsurface
theinand
three-layer
powder
peak
and
644capsule.
cm−1(at
capsule
peak
at each
offset
distance
and
the−1of
trend
ratio
outweighed
that the
ofdistance
the
Figure
8 shows
the ratio
peak
intensities
1009 cm
(urea)
capsule-urea
The
ratioindicates
values
increased
significantly,
2 powder
aton0the
mm
offset
values.
A−1sample.
higher
ratio
that
the signal
intensity
of from
the subsurface
component
644 cm
(capsule)
as value
a function
of offset
distance
(at
0.5
mm intervals
upabout
to 3 mm)
surface
of to 15 at
the three-layer
sample.
The
ratio values
increased
significantly,
from
about
0outweighed
mmand
3 mmoutweighed
offset,
indicating
thatcapsule.
the spectral
contribution
from
subsurface
gradually
that capsule-urea
of the
Figure
8 shows
the ratio
ofthe
peak
intensitiesurea
at 1009
cm2−1at
(urea)
−1 to
offset
15
at
3
mm
offset,
indicating
that
the
spectral
contribution
from
the
subsurface
urea
644
cm
(capsule)
as
a
function
of
offset
distance
(at
0.5
mm
intervals
up
to
3
mm)
on
the
surface
of that,
the spectral contribution from the capsule layer. The increasing intensity ratio trend shows
gradually
outweighed
the spectral
contribution
from the
capsule layer.
The increasing
intensity
ratio
the
three-layer
capsule-urea
sample.
The
ratio
values
increased
significantly,
from
about
2
at
0
mm
at increased offset distance, the incident laser penetrates deeper through the sample in retrieving
trend
shows
that,
at offset,
increased
offset distance,
the incidentcontribution
laser penetrates
through the
offset
to
15 at
3 mm
that the
fromdeeper
the subsurface
urea
subsurface
layer
information.
A indicating
similar pattern
wasspectral
found for all samples.
sample in
retrieving subsurface
layer
information.
A similar
pattern
wasThe
found
for all samples.
gradually
outweighed
the spectral
contribution
from
the capsule
layer.
increasing
intensity ratio
trend shows that, at increased offset distance, the incident laser penetrates deeper through the
sample in retrieving subsurface layer information. A similar pattern was found for all samples.
Figure 8. Ratio of Raman peak intensities at 1009 cm−1 (urea) and 644 cm−1 (capsule) for offset distance
Figure 8.
Ratio of Raman peak intensities at 1009 cm−1 (urea) and 644 cm−1 (capsule) for offset distance
up to 3 mm on the three-layer capsule-urea sample.
up to 3 mm on the three-layer capsule-urea sample.
3.3.
Identification
of Chemical
Components
Figure
8. Ratio of
Raman peak
intensities at 1009 cm−1 (urea) and 644 cm−1 (capsule) for offset distance
3.3. Identification
of on
Chemical
Components
upSORS
to 3 mm
the three-layer
capsule-urea
sample.
data
measured
for offset
distances
of 0 to 10 mm for each sample were resolved into pure
component spectra by SMA. Figure 9 shows the pure component spectra obtained for the one-,
SORS
data measured
forComponents
offset distances of 0 to 10 mm for each sample were resolved into
3.3.
Identification
of Chemical
three-,
five-, and
eight-layer
samples, as well as the reference spectra used to identify the resolved
pure component
spectra
by SMA.
Figure
9 shows
the9apure
component
spectraofobtained
for the one-,
pure
component
spectra
based
on
SID
values.
Figure
shows
spectral
features
urea that
were
SORS
data measured
for offset
distances
of
0 to 10
mm forthe
each
sample
were resolved
into
pure
three-, five-,
andretrieved
eight-layer
samples,
well asurea
the spectrum.
referenceThe
spectra
used
identifyfor
the
effectively
by matching
theas
reference
resolved
ureatospectrum
theresolved
component spectra by SMA. Figure 9 shows the pure component spectra obtained for the one-,
eight-layer sample
showed
spectral
distortion
between
630 cm−1the
to 660
cm−1 that
might beof
due
pure three-,
component
spectra
basedsome
on SID
values.
Figure
9a shows
spectral
features
urea that
five-, and eight-layer samples, as well as the reference spectra used to identify the resolved
to
the
increased
capsule
wall
thickness.
Figure
9b
shows
that
the
resolved
capsule
spectra
recovered
werepure
effectively
retrieved
matching
reference
urea
spectrum.
Thefeatures
resolved
ureathat
spectrum
component
spectraby
based
on SID the
values.
Figure 9a
shows
the spectral
of urea
were for
almost all of the spectral features present in the reference capsule spectrum. The
spectra
for
−1resolved
−1 that
the eight-layer
sample
showed
some
spectral
distortion
between
630
cm
to
660
cm
effectively retrieved by matching the reference urea spectrum. The resolved urea spectrum for themight
ibuprofen and acetaminophen in Figure 9c,d, respectively, also show features well matched to their
−1 to 660
−1
be due
to the increased
capsule
wall
thickness.
Figure
9bofshows
that
thecm
resolved
capsule
spectra
eight-layer
showed
some
spectral
distortion
between
630
cmprominent
might
respectivesample
reference
spectra.
The
retrieval
and matching
very
keythat
peaks
for be
thedue
to the
increased
capsule
wall
thickness.
Figure
9b
shows
that
the
resolved
capsule
spectra
recovered
recovered
almost
all
of
the
spectral
features
present
in
the
reference
capsule
spectrum.
The
resolved
−1
−1
−1
−1
components (at 1009 cm for urea, 833 cm for ibuprofen, 859 cm for acetaminophen, and 644 cm
almost
all of the
spectral
features
the reference
capsule
The
spectra
for
spectra
forgelatin
ibuprofen
andshows
acetaminophen
ininFigure
9c,d,
respectively,
also show
features
wellofmatched
for
capsule)
that present
SORS coupled
with
SMA
can be spectrum.
effectively
usedresolved
for detection
ibuprofen
and
acetaminophen
in
Figure
9c,d,
respectively,
also
show
features
well
matched
to
their
encapsulated
chemicals
and
for
identification
of
each
chemical
component.
to their respective reference spectra. The retrieval and matching of very prominent key peaks for
respective reference spectra.
The retrieval and−1matching of very prominent
peaks for the
the components
(at 1009 cm−1 for
urea, 833 cm for ibuprofen, 859 cm−1 forkey
acetaminophen,
and
−1 for urea, 833 cm−1 for ibuprofen, 859 cm−1 for acetaminophen, and 644 cm−1
components
(at
1009
cm
644 cm−1 for gelatin capsule) shows that SORS coupled with SMA can be effectively used for detection
for gelatin capsule) shows that SORS coupled with SMA can be effectively used for detection of
of encapsulated
chemicals and for identification of each chemical component.
encapsulated chemicals and for identification of each chemical component.
(a)
(b)
(a)
(b)
Figure 9. Cont.
Sensors 2017, 17, 618
10 of 12
Sensors 2017, 17, 618
10 of 12
(c)
(d)
Figure 9. Pure component spectra of (a) urea; (b) capsule; (c) ibuprofen; and (d) acetaminophen
Figure 9. Pure component spectra of (a) urea; (b) capsule; (c) ibuprofen; and (d) acetaminophen
resolved from mixed-contribution SORS spectra and identified by SID values.
resolved from mixed-contribution SORS spectra and identified by SID values.
4. Conclusions
4. Conclusions
This study developed a spatially offset Raman spectroscopy (SORS) method to retrieve Raman
information
of encapsulated
samplesoffset
of urea,
ibuprofen,
and acetaminophen
powders
prepared
This study developed
a spatially
Raman
spectroscopy
(SORS) method
to retrieve
Raman
using upof
toencapsulated
eight layers of samples
gelatin capsule.
SORS
spectra collected
over a spatialpowders
offset range
of 0 (nousing
information
of urea,
ibuprofen,
and acetaminophen
prepared
10 mm
mixed spectral
information
of the over
capsules
and chemicals.
Keyof
chemical
up tooffset)
eightto
layers
ofcontained
gelatin capsule.
SORS spectra
collected
a spatial
offset range
0 (no offset)
peaks that initially appeared to be nonexistent in spectra measured at no offset were gradually
to 10 mm contained mixed spectral information of the capsules and chemicals. Key chemical peaks
enhanced with increased offset distances to become more readily apparent. On the other hand,
that initially appeared to be nonexistent in spectra measured at no offset were gradually enhanced
capsule peak intensity was attenuated by greater offset distance. By conducting spectral
withmeasurements
increased offset
distancesoffset
to become
more
readily apparent.spectra
On the
other
hand,that
capsule
at different
distances,
mixed-contribution
are
obtained
have peak
intensity
was
attenuated
by
greater
offset
distance.
By
conducting
spectral
measurements
at
different
differences in contribution from the upper surface layers and the deeper subsurface areas at different
offsetoffset
distances,
mixed-contribution
spectra
are
obtained
that have components.
differences in contribution from
distances,
allowing for detection
of both
surface
and subsurface
the upperFor
surface
layers and
the deeper
subsurface
at different
offset distances,
allowing for
identification
of individual
components
in areas
each sample,
self-modeling
mixture analysis
(SMA)ofmethod
was used
to subsurface
decompose the
mixed spectral information in the SORS data and extract
detection
both surface
and
components.
pure
component spectra
of capsule
and individual
Almost all spectral
For identification
of individual
components
in each chemicals.
sample, self-modeling
mixturefeatures
analysisof(SMA)
capsules
and
chemicals
were
retrieved
in
pure
component
spectra
that
matched
the
method was used to decompose the mixed spectral information in the SORS data andreference
extract pure
spectrum of individual components. The corresponding components for each retrieved pure
component spectra of capsule and individual chemicals. Almost all spectral features of capsules
component spectra were identified by spectral information divergence (SID). Despite ibuprofen and
and chemicals were retrieved in pure component spectra that matched the reference spectrum of
acetaminophen having some similar spectral features with Raman peaks in close proximity, spectra
individual
corresponding
components
for each identified.
retrieved pure
component
spectra
for the components.
two chemicalsThe
were
effectively extracted
and accurately
This shows
that the
weremethod
identified
by spectral
(SID).toDespite
and
acetaminophen
having
presented
in thisinformation
study can bedivergence
effectively used
retrieve ibuprofen
information
from
layered samples
someand
similar
spectral
features
withcomponents
Raman peaks
insample.
close proximity, spectra for the two chemicals were
identify
different
chemical
in the
effectively
extracted
and accurately
identified.
Thistoshows
that
the method
presented
in this
This
study demonstrated
an effective
method
retrieve
information
of layered
samples
bystudy
SORS
measurement,
extract
pure
component
spectra
of
individual
components
in
mixed
spectral
can be effectively used to retrieve information from layered samples and identify different chemical
data by SMA
method,
and finally identify different chemicals in the mixture using SID. The wider
components
in the
sample.
applications
of
this
method
caneffective
be detection
of concealed
drugs,
identification
of materials
inside
This study demonstrated an
method
to retrieve
information
of layered
samples
by SORS
coated containers, quality control of capsules, analysis of constituents inside packaged food, and
measurement, extract pure component spectra of individual components in mixed spectral data by
evaluating internal food safety and quality attributes.
SMA method, and finally identify different chemicals in the mixture using SID. The wider applications
of this
method
can be detection
of concealed
drugs,conceived
identification
of materials
inside coated
Author
Contributions:
Kuanglin Chao
and Sagar Dhakal
and designed
the experiments;
Kuanglincontainers,
Chao
andcontrol
Sagar Dhakal
performed
the experiments;
Jianweiinside
Qin, Yankun
Peng,food,
and Moon
S. Kim helped
in the food
quality
of capsules,
analysis
of constituents
packaged
and evaluating
internal
experiment;
Kuanglin
Chao, Sagar Dhakal, and Walter F. Schmidt analyzed the data; Kuanglin Chao,
safety
and quality
attributes.
Sagar Dhakal, and Diane E. Chan wrote the paper.
Author
Contributions:
Kuanglin
Chao noand
Sagar
Dhakal conceived and designed the experiments;
Conflicts
of Interest: The
authors declare
conflict
of interest.
Kuanglin Chao and Sagar Dhakal performed the experiments; Jianwei Qin, Yankun Peng, and Moon S. Kim
helped
in the experiment; Kuanglin Chao, Sagar Dhakal, and Walter F. Schmidt analyzed the data; Kuanglin Chao,
References
Sagar Dhakal, and Diane E. Chan wrote the paper.
1.
Yang, D.; Ying, Y. Applications of Raman spectroscopy in agricultural products and food analysis: A
review. Appl. Spectrosc. Rev. 2011, 46, 539–560.
Conflicts of Interest: The authors declare no conflict of interest.
Sensors 2017, 17, 618
11 of 12
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Yang, D.; Ying, Y. Applications of Raman spectroscopy in agricultural products and food analysis: A review.
Appl. Spectrosc. Rev. 2011, 46, 539–560. [CrossRef]
Eliasson, C.; Matousek, P. Noninvasive authentication of pharmaceutical products through packaging using
spatially offset Raman spectroscopy. Anal. Chem. 2007, 79, 1696–1701. [CrossRef] [PubMed]
Dhakal, S.; Li, Y.; Peng, Y.; Chao, K.; Qin, J.; Guo, L. Prototype instrument development for non-destructive
detection of pesticide residue in apple surface using Raman technology. J. Food Eng. 2014, 123, 94–103.
[CrossRef]
Dhakal, S.; Chao, K.; Schmidt, W.; Qin, J.; Kim, M.; Chan, D. Evaluation of turmeric powder adulterated
with metanil yellow using FT-Raman and FT-IR spectroscopy. Foods 2016, 5, 36. [CrossRef] [PubMed]
Hajjou, M.; Qin, Y.; Bradby, S.; Bempong, D.; Lukulay, P. Assessment of the performance of a handheld
Raman device for potential use as a screening tool in evaluating medicines quality. J. Pharm. Biomed. Anal.
2013, 74, 47–55. [CrossRef] [PubMed]
Izake, E.L. Forensic and homeland security applications of modern portable Raman spectroscopy.
Forensic Sci. Int. 2010, 202, 1–8. [CrossRef] [PubMed]
Krafft, C.; Sergo, V. Biomedical applications of Raman and infrared spectroscopy to diagnose tissues.
Spectroscopy 2006, 20, 195–218. [CrossRef]
Matousek, P.; Parker, A.W. Non-invasive probing of pharmaceutical capsules using transmission Raman
spectroscopy. J. Raman Spectrosc. 2007, 38, 563–567. [CrossRef]
Eliasson, C.; Macleod, N.A.; Jayes, L.C.; Clarke, F.C.; Hammond, S.V.; Smith, M.R.; Matousek, P. Non-invasive
quantitative assessment of the content of pharmaceutical capsules using transmission Raman spectroscopy.
J. Pharm. Biomed. Anal. 2008, 47, 221–229. [CrossRef] [PubMed]
Johansson, J.; Sparen, A.; Svensson, O.; Folestad, S.; Claybourn, M. Quantitative transmission Raman
spectroscopy of pharmaceutical tablets and capsules. Appl. Spectrosc. 2007, 61, 1211–1218. [CrossRef]
[PubMed]
Shin, K.; Chung, H.; Kwak, C.-W. Transmission Raman measurement directly through packed corn kernels
to improve sample representation and accuracy of compositional analysis. Analyst 2012, 137, 3690–3696.
[CrossRef] [PubMed]
Schulmerich, M.V.; Walsh, M.J.; Gelbert, M.K.; Kong, R.; Kole, M.R.; Harrison, K.; McKinney, J.; Thompson, D.;
Kull, L.S.; Bhargava, R. Protein and oil composition predictions of single soybeans by transmission Raman
spectroscopy. J. Agric. Food. Chem. 2012, 60, 8097–8102. [CrossRef] [PubMed]
Matousek, P.; Clark, I.P.; Draper, E.R.C.; Morris, M.D.; Goodship, A.E.; Everall, N.; Towrie, M.; Finney, W.F.;
Parker, A.W. Subsurface probing in diffusely scattering media using spatially offset Raman spectroscopy.
Appl. Spectrosc. 2005, 59, 393–400. [CrossRef] [PubMed]
Olds, W.J.; Jaatinen, E.; Fredericks, P.; Cletus, B.; Penayioutou, H.; Izake, E.L. Spatially offset Raman
spectroscopy (SORS) for the analysis and detection of packaged pharmaceuticals and concealed drugs.
Forensic Sci. Int. 2011, 212, 69–77. [CrossRef] [PubMed]
Stone, N.; Baker, R.; Rogers, K.; Parker, A.W.; Matousek, P. Subsurface probing of calcifications with spatially
offset Raman spectroscopy (SORS): Future possibilities for the diagnosis of breast cancer. Analyst 2007, 132,
899–905. [CrossRef] [PubMed]
Keller, M.D.; Majumder, S.K.; Mahadevan-Jansen, A. Spatially offset Raman spectroscopy of layered soft
tissues. Opt. Lett. 2009, 34, 926–928. [CrossRef] [PubMed]
Afseth, N.K.; Bloomfield, M.; Wold, J.P.; Matousek, P. A novel approach for subsurface through-skin analysis
of salmon using spatially offset Raman spectroscopy (SORS). Appl. Spectrosc. 2014, 68, 255–262. [CrossRef]
[PubMed]
Qin, J.; Chao, K.; Kim, M. Nondestructive evaluation of internal maturity of tomatoes using spatially offset
Raman spectroscopy. Postharvest Biol. Technol. 2012, 71, 21–31. [CrossRef]
Qin, J.; Kim, M.; Schmidt, W.; Cho, B.-K.; Peng, Y.; Chao, K. A line-scan hyperspectral Raman system for
spatially offset Raman spectroscopy. J. Raman Spectrosc. 2016, 47, 437–443. [CrossRef]
Hargreaves, M.D.; Macleod, N.A.; Brewster, V.L.; Munshi, T.; Edwards, H.G.M.; Matousek, P. Application of
portable Raman spectroscopy and benchtop spatially offset Raman spectroscopy to interrogate concealed
biomaterials. J. Raman Spectrosc. 2009, 40, 1875–1880. [CrossRef]
Sensors 2017, 17, 618
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
12 of 12
Lodder, R.A.; Selby, M.; Hieftje, G.M. Detection of capsule tampering by near-infrared reflectance analysis.
Anal. Chem. 1992, 59, 1921–1930. [CrossRef]
Lanzarotta, A.; Crowe, J.B.; Witkowski, M.; Gamble, B.M. A multidisciplinary approach for the analysis of an
adulterated dietary supplement where the active pharmaceutical ingredient was embedded in the capsule
shell. J. Pharm. Biomed. Anal. 2012, 67–68, 22–27. [CrossRef] [PubMed]
Dhakal, S.; Chao, K.; Qin, J.; Kim, M.; Chan, D. Raman spectral imaging for quantitative contaminant
evaluation in skim milk powder. J. Food Measur. Charact. 2016, 10, 374–386. [CrossRef]
Winding, W.; Guilment, J. Interactive self- modeling mixture analysis. Anal. Chem. 1991, 63, 1425–1432.
[CrossRef]
Winding, W.; Gallagher, N.B.; Shaver, J.M.; Wise, B.M. A new approach for interactive self-modeling mixture
analysis. Chemom. Intell. Lab. Syst. 2005, 77, 85–96. [CrossRef]
Qin, J.; Chao, K.; Kim, M. Simultaneous detection of multiple adulterants in dry milk using macro-scale
Raman chemical imaging. Food Chem. 2013, 138, 998–1007. [CrossRef] [PubMed]
Dhakal, S.; Chao, K.; Qin, J.; Kim, M.; Peng, Y.; Chan, D. Identification and evaluation of composition in food
powder using point-scan Raman spectral imaging. Appl. Sci. 2017, 7, 1. [CrossRef]
Batonneau, Y.; Laureyns, J.; Merlin, J.-C.; Bremard, C. Self-modeling mixture analysis of Raman
microspectrometric investigations of dust emitted by lead and zinc smelters. Anal. Chim. Acta 2001,
446, 23–37. [CrossRef]
Chang, C.I. An information theoretic-based approach to spectral variability, similarity and discriminability
for hyperspectral image analysis. IEEE Trans. Inf. Theory 2000, 46, 1927–1932. [CrossRef]
Liapis, K.; Jayasooriya, U.A.; Kettle, S.F.A. Crystalline Urea: Evidence for vibrational delocalization.
J. Phys. Chem. 1985, 89, 4560–4565. [CrossRef]
Schmidt, W.F.; Broadhurst, C.L.; Qin, J.; Li, H.; Nguyen, J.K.; Chao, K.; Hapeman, C.J.; Shelton, D.R.; Kim, M.S.
Continuous temperature- dependent Raman spectroscopy of melamine and structural analogs detection in
milk powder. Appl. Spectrosc. 2015, 69, 398–406. [CrossRef] [PubMed]
Jubert, A.; Legarto, M.L.; Massa, N.E.; Tevez, L.L.; Okulik, N.B. Vibrational and theoretical studies
of non- steroidal anti- inflammatory drugs ibuprofen [2-(4-isobutylphenyl) propionic acid]; Naproxen
[6- methoxy-α-methyl-2-napthalene acetic acid] and Tolmetin acids [1-methyl-5-(4-methylbenzoyl)1H-pyrrole-2-acetic acid]. J. Mol. Struct. 2006, 783, 34–51.
Andronie, L.; Miresan, V.; Raducu, C.; Pop, I.; Coroian, A.; Coroian, C.O. The adsorption behavior of buffered
aspirin monitored by Raman and surface-enhanced Raman spectroscopy. HVM Bioflux 2014, 6, 187–190.
Pestaner, J.P.; Mullick, F.G.; Centeno, J.A. Characterization of acetaminophen: Molecular microanalysis with
Raman microprobe spectroscopy. J. Forensic Sci. 1996, 41, 1060–1063. [CrossRef] [PubMed]
Abrahams, J.P.; Leslie, A.G.W.; Lutter, R.; Walker, J.E. Structure at 2.4A resolution of F1 - ATPase from bovine
heart mitochondria. Nature 1994, 370, 621–628. [CrossRef] [PubMed]
Maiti, N.C.; Apetri, M.M.; Zagorski, M.G.; Carey, P.R.; Anderson, V.E. Raman spectroscopic characterization
of secondary structure in natively unfolded proteins: α-synuclein. J. Am. Chem. Soc. 2004, 126, 2399–2408.
[CrossRef] [PubMed]
© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).