Isotope edited internal standard method for quantitative surface-enhanced raman spectroscopy

Article abstract of DOI:10.1021/ac050338h

A new isotope edited internal standard (IEIS) method for quantitative surface-enhanced Raman spectroscopy (SERS) is demonstrated using rhodamine 6G (R6G-d₀) and rhodamine 6G (R6G-d₄) edited with deuterium. The reproducibility and accuracy of the IEIS method is investigated both under optical resonance (SERRS) and non-resonance (SERS) conditions. A batch-to-batch concentration measurement reproducibility of better than 3% is demonstrated over a concentration range of 200 pM-2 μM with up to a factor of 3 difference between the concentration of the analyte and its IEIS. The superior performance of the IEIS method is further illustrated by comparing results obtained using absolute SERS/SERRS intensify calibration (with no internal standard) or using adenine (rather than R6G-d₄) as an internal standard for R6G concentration quantization. Potential biomedical gene expression and comparative proteomic applications of the IEIS method are discussed.

Full text of DOI:10.1021/ac050338h

Anal. Chem. 2005, 77, 3563-3569
Isotope Edited Internal Standard Method for
Quantitative Surface-Enhanced Raman
Spectroscopy
Dongmao Zhang,*,† Yong Xie, Shirshendu K. Deb, V. Jo Davison, and Dor Ben-Amotz†
†
‡
‡,§
Department of Chemistry, Department of Medicinal Chemistry and Molecular Pharmacology, and The Bindley Bioscience
Center, Purdue Discovery Park, Purdue University, West Lafayette, Indiana 47907
A new isotope edited internal standard (IEIS) method for
quantitative surface-enhanced Raman spectroscopy (SERS)
challenge because of (1) the difficulties associated with the
production of reproducible SERS active substrates, (2) the
complex characteristics (both spatial and temporal) of the SERS
substrates, (3) the strong dependence of the SERS enhancement
on the distance between the analyte and the SERS substrates,
and (4) variations of SERS enhancement with the surface coverage
is demonstrated using rhodamine 6G (R6G-d
0
) and
rhodamine 6G (R6G-d ) edited with deuterium. The
4
1
3
reproducibility and accuracy of the IEIS method is inves-
tigated both under optical resonance (SERRS) and non-
resonance (SERS) conditions. A batch-to-batch concen-
tration measurement reproducibility of better than 3% is
demonstrated over a concentration range of 200 pM-2
µM with up to a factor of 3 difference between the
concentration of the analyte and its IEIS. The superior
performance of the IEIS method is further illustrated by
comparing results obtained using absolute SERS/SERRS
intensity calibration (with no internal standard) or using
of the analyte on the substrate (related to the distribution of SERS
active hot spots).1
3,14
In addition, quantitative concentration
measurements using optical methods (including SERS as well as
normal Raman or fluorescence) must contend with intensity
variations produced by changes in excitation, collection efficiency,
or both. Correcting for such variations is most often accomplished
using either an internal or an external standard to calibrate the
correlation between the optical signal and the concentration (or
amount) for the analyte of interest. Here we describe a new
isotopically edited internal standard (IEIS) method that may be
used for quantitative SERS/SERRS measurements over a wide
concentration range with unprecedented accuracy and reproduc-
ibility.
4
adenine (rather than R6G-d ) as an internal standard for
R6G concentration quantization. Potential biomedical
gene expression and comparative proteomic applications
of the IEIS method are discussed.
Since the discovery of the surface-enhanced Raman scattering
The IEIS method differs in important ways from the following
previously proposed methods for improving the accuracy of
quantitative SERS measurements. For example, Smith et al.
employed a flow cell device for in situ aggregation of Ag colloidal
fabricated with the Lee-Misel method1 and found good linearity
and reproducibility when using the same batch of a SERS active
colloidal solution. However, when different batches of colloidal
solutions were used, the reproducibility of the SERRS signal with
mitoxantrone concentration deteriorated significantly (with cali-
bration slope differences of up to 60%, even though all the other
1
(
SERS), various SERS active substrates and molecules have been
6
reported with a typical Raman signal enhancement of 10 . Under
electronic resonance (SERRS) conditions, far greater enhance-
ments may be attained, and single-molecule detection limits have
been reported for rhodamine 6G, adenine, crystal violet, and other
5,16
2
-5
SERRS active molecules. Because of the high sensitivity of the
SERS/SERRS techniques and high information content of the
resulting vibrational spectra, SERS active molecules have been
employed as labeling reagents for bioanalytical applications which
-
18
15
enabled detection of attomole (10 mol) quantities of proteins
experiment conditions remained the same ). More recently, an
internal standard method was proposed to improve the accuracy
-
15
6-12
or DNAs down to femtomolar (10 mol/L) concentrations.
However, accurate quantitative analysis with SERS remains a
(7) Faulds, K.; Barbagallo, R. P.; Keer, J. T.; Smith, W. E.; Graham, D. Analyst
*
To whom correspondence should be addressed. E-mail: zhang17@purdue.edu.
Department of Chemistry.
Department of Medicinal Chemistry and Molecular Pharmacology.
The Bindley Bioscience Center.
2004, 129, 567-568.
†
(8) Faulds, K.; Smith, W. E.; Graham, D. Anal. Chem. 2004, 76, 412-417.
(9) Graham, D.; Brown, R.; Smith, W. E. Chem. Commun. 2001, 1002-1003.
(10) Culha, M.; Stokes, D.; Allain, L. R.; Vo-Dinh, T. Anal. Chem. 2003, 75,
6196-6201.
‡
§
(
(
(
1) Fleischmann, M.; Hendra, P. J.; McQuillan, A. J. Chem. Phys. Lett. 1974,
6, 163.
2
(11) Graham, D.; Mallinder, B. J.; Whitcombe, D.; Watson, N. D.; Smith, W. E.
2) Kneipp, K.; Kneipp, H.; Kartha, V. B.; Manoharan, R.; Deinum, G.; Itzkan,
I.; Dasari, R. R.; Feld, M. S. Phys. Rev. E 1998, 57, R6281-R6284.
3) Koo, T.-W.; Chan, S.; Sun, L.; Su, X.; Zhang, J.; Berlin, A. A. Appl. Spectrosc.
Anal. Chem. 2002, 74, 1069-1074.
(12) Cao, Y. C.; Jin, R.; Nam, J.-M.; Thaxton, C. S.; Mirkin, C. A. J. Am. Chem.
Soc. 2003, 125, 14676-14677.
2
004, 58, 1401-1407.
(13) Lacy, W. B.; Olson, L. G.; Harris, J. M. Anal. Chem. 1999, 71, 2564-2570.
(14) Campion, A.; Kambhampati, P. Chem. Soc. Rev. 1998, 27, 241-250.
(15) McLaughlin, C.; MacMillan, D.; McCardle, C.; Smith, W. E. Anal. Chem.
2002, 74, 3160-3167.
(
(
4) Nie, S.; Emory, S. R. Science 1997, 275, 1102-1106.
5) Kneipp, K.; Wang, Y.; Kneipp, H.; Perelman, L. T.; Itzkan, I.; Dasari, R. R.;
Feld, M. S. Phys. Rev. Lett. 1997, 78, 1667-1670.
(
6) Cao, Y. C.; Jin, R.; Mirkin, C. A. Science 2002, 297, 1536-1540.
(16) Lee, P. C.; Meisel, D. J. Phys. Chem. 1982, 86, 3391-3395.
1
0.1021/ac050338h CCC: $30.25 © 2005 American Chemical Society
Analytical Chemistry, Vol. 77, No. 11, June 1, 2005 3563
Published on Web 04/26/2005

for SERS (colloidal) quantification by using the SERS signal
generated from a self-assembled monolayer (SAM) as an internal
be sufficiently different to facilitate independent measurement of
their SERS intensities in a mixture.
1
7
standard. With this method, the high coverage of the SAM is
presumed to prevent chemisorption of the analyte onto the SERS
active surfaces and thus to improve reproducibility.17 However,
this approach also has intrinsic limitations. For example, because
of the sharp drop-off of the SERS enhancement with the distance
Given the above consideration, one might expect that an ideal
internal standard for any analyte would be an isotopically edited
version of the analyte of interest. In fact, isotope editing is a
commonly used technique in vibrational spectroscopy, to aid in
assignment of spectroscopic features associated with specific
functional groups. Advantages of isotopic editing as an internal
standard method include the fact that the two compounds are
expected to have (a) virtually the same chemical and physical
properties but (b) a readily measurable and quantifiable spectro-
scopic differences. The fact that isotope editing can produce
significant spectral changes is illustrated, for example, by the
14
between the analyte and SERS surface, the limit of the detection
and the dynamic range with the SAM approach has been severely
compromised (because of the greater distance between the analyte
and SERS surface created by the SAM coating). Furthermore, the
different local environments around the SAM and the analyte
molecules may produce a different response to experiment
parameters such as laser intensity and frequency. These and other
factors may explain the relatively large prediction errors (root
mean prediction error of 0.5 µM for samples between 0.1 and 5
µM) observed when using this SAM internal standard method.17
The present IEIS method overcomes the limitations of previous
methods because the analyte and standard molecules have
virtually identical chemical properties and so their relative SERS/
SERRS intensity is far less sensitive to batch-to-batch colloid
solution variations and optical excitation/collection parameters (as
further described below).
-
1
observed peak red-shifts of 50 and 30 cm produced by the
2
vibration of the ring breathing mode of benzene produced H (D)
or C13 editing, respectively,
18,19
which are quite significant given
that the corresponding Raman band has a full width at half-
-
1
maximum of <6 cm . Furthermore, since most molecules with
large SERS/SERRS activities contain aromatic functional groups
and the Raman signals of these functional groups are in general
the most prominent features in the resulting Raman spectra,
isotopic edition of aromatic groups should provide a widely
applicable IEIS method,
In general, the SERS signal intensity, I, can be a function of
Like other internal standard methods, SERS quantification with
IEIS is carried out by mixing the sample of interest with its IEIS
of known concentration before incubation of the mixture with a
SERS active substrate (e.g., a colloidal solution). After SERS
acquisition, the concentration ratio of the analyte and the internal
standard may thus be determined from the ratio of the spectral
features associated with the two compounds.
δ
several variables as shown in eq 1, where
I
is the signal
κ
δ
s
I ) C I R
κ κ k
(1)
κ
intensity obtained from sample κ of concentration of C
κ
, under a
s
given set of experiment conditions, δ, while
I
represents the
κ
The SERS and SERRS spectroscopy of rhodamine 6G (R6G)
signal intensity obtained with unit analyte concentration under a
given set of standard conditions, s, and R represents the relative
SERS enhancement factor, which may in general also be a function
of C , as well as the characteristics of the SERS substrate and
has been extensively studied,4
,20,21
as it is one of the most
κ
commonly used SERS/SERRS tags in DNA and protein detections
7,8,10,22
applications,
as well as in SERRS single-molecule detection
κ
4
studies. Thus, R6G was selected as the model compound to
demonstrate the feasibility and performance of IEIS for SERRS
and SERS quantitative analysis. In addition, to investigate the
feasibility of using molecules other than IEIS, we have also
investigated the SERS spectra of mixtures containing both R6G
and adenine. The latter studies serve to clearly demonstrate the
superior performance of the IEIS method as opposed to the use
of chemically different analyte and internal standard compounds.
The robustness and accuracy of SERS/SERRS with IEIS was
evaluated using several batches of colloidal solution, with R6G
concentrations varying from 200 pM to 2 µM. Potential biomedical
applications with IEIS are discussed.
Raman system.
For an internal standard to be effective, all the SERS intensity
variations introduced by anything other than sample concentration
must to be compensated by internal SERS intensity standard. In
other words, an ideal internal standard method is one for which
the SERS intensity ratio of the analyte, a, and internal standard
reference, r, are strictly proportional to the ratio of their corre-
sponding concentrations.
δ
C Is
I
I
a
a a
)
(2)
δ
r
C Is
r r
EXPERIMENTAL SECTION
In other words, the relative enhancement factors of the analyte
and reference compound must be identical, R ) R , and this
Chemicals. All the reagents used for organic synthesizing and
a
r
colloidal solution and adenine were of analytical grade (Sigma-
Aldrich), High-purity water (Millipore) was used throughout this
study.
equality should not depend on either the absolute or relative
concentrations of the analyte and reference compounds or any
other experimental variables. This implies that the ideal internal
standard should have chemical properties that are as similar as
possible to the analyte of interest. However, to be able to
differentiate and quantify the spectral contribution for the analyte
and its internal standards, their SERS spectral features have to
(18) Shimanouchi, T. Tables of Molecular Vibrational Frequencies; National Bureau
of Standards: Washington, DC. 1972.
(
(
19) Painter, P. C.; Koenig, J. L. Spectrochim. Acta 1977, 33A, 1003-1018.
20) Bosnick, K. A.; Jiang, J.; Brus, L. E. J. Phys. Chem. B 2002, 106, 8096-
8099.
(21) Li, G.; Li, H.; Mo, Y.; Huang, X.; Chen, L. Chem. Phys. Lett. 2000, 330,
(
17) Loren, A.; Engelbrektsson, J.; Eliasson, C.; Josefson, M.; Abrahamsson, J.;
249-254.
Johansson, M.; Abrahamsson, K. Anal. Chem. 2004, 76, 7391-7395.
(22) Graham, D.; Fruk, L.; Smith, W. E. Analyst 2003, 128, 692-699.
3564 Analytical Chemistry, Vol. 77, No. 11, June 1, 2005

Table 1. Integration Times for SERS and SERRS
Measurements
excitation
laser
total concentration of
silver
colloidal
batch
R6G-d0 and R6G-d4
nm mW 200 pM 2 nM 20 nM 200 nM 2 µM
1
2
3
4
5
514
514
514
633
633
10
10
10
6
30 s
30 s
30 s
5 s
5 s
5 s
500 ms 50 ms
500 ms 50 ms
500 ms 10 ms
5 s
5 s
1 s
1 s
200 ms
100 ms
6
Figure 1. Schematic diagram representing the synthetic procedure
used to produce rhodamine 6G (R6G) with no deuterium (R6G-d0)
and with four deuterium substitutions (R6G-d4).
SERRS and SERS measurement are listed in Table 1; no attempt
was made to optimize the signal intensities for each measurement.
Spectral Data Analysis. for all the spectral analysis, the
following simple least-squares spectral decomposition algorithm
was developed and implemented using Matlab (MathWorks Inc.)
Silver Colloid. Silver colloidal solution was synthesized
according to the Lee-Misel method by citrate reduction of silver
1
6
to calculate the relative concentration ratio of R6G-d
Let S and S stand for the SERS/SERRS spectra obtained with
pure R6G-d and R6G-d , respectively, and D for the SERS/SERRS
spectrum measured with mixture of an unknown quantity of R6G-
with R6G-d of concentration C . Since spectrum D contains
only spectral contribution from S and S plus experiment noise,
0 4
and R6G-d .
nitrate. Five batches of colloidal solution were synthesized
independently and the first three batches were used for SERRS
measurement and the last two for SERS. The aging period for
different batches of colloidal solution varied from 2 h to 4 days.
Instruments. The SERS spectra were obtained using a home-
built micro-Raman system with a 632.8-nm HeNe laser (with 10
mW at the sample), while the SERRS measurements were
performed with another home-built Raman system with 514-nm
argon ion excitation lasers (with 6 mW at the sample). With both
systems, the back-reflected Raman signal was collected using a
1
2
0
4
d
0
4
r
1
2
spectrum D can be decomposed into spectra S
eq 3 with the least-squares method:
1
and S
2
by solving
S
1
|
C C |
) D
(3)
1
2
| |
S
2
2
0× Olympus objective and coupled to a spectrograph with a fiber
bundle for detection with liquid nitrogen-cooled CCD detected.
The spectrographs used on both systems have focal lengths of
However, it should be noticed that C
1
and C
2
are not the
0
.3 m (Acton, SpectraPro 300i) and 1200 grids/mm gratings.
concentrations of S
their relative contributions to mixture spectrum D before a proper
adjustment of spectral intensity of S or S . The goal for this
1 2
and S ; in fact, they do not even represent
Synthesis of the R6G and Its IEIS. R6G and its IEIS
derivative were synthesized by coupling 3-(ethylamino)-4-meth-
ylphenol with commercially available phthalic anhydride and
1
2
adjustment is to make the intensity ratio of the component spectra
equal to what would be obtained with the component spectra each
acquired under exactly the same conditions. This can be done by
4
phthalic-d anhydride, respectively, followed by ethylation of the
free carboxylic acid groups. The synthetic route is illustrated in
Figure 1. Since four H atoms are substituted with D in the
isotopically edited R6G, the two compounds will from hereon be
1 2 1
simply finding a multiplying constant for S or S so that the C /
C
2
ratio determined with the adjusted component spectra matrix
0 4
abbreviated as R6G-d and R6G-d , while the term R6G will
will be equal to 1 for any SERS/SERRS spectra obtained with the
mixture consisting of exactly 50% of each component, and the
resulting intensity-calibrated component spectra are used for all
subsequent spectral analysis.
continue to be used to refer to either one or both of the isotopes.
Quantitative Studies. All the SERRS experiments with R6G
were repeated three times, each with a different batch of silver
colloidal solution. The samples used for each repeated trial were
the same: the total R6G concentration varied from 200 pM to
RESULT AND DISCUSSION
2
00 nM; quantitative SERRS analysis with IEIS was only carried
out at two concentrations of 200 pM and 200 nM. For the SERS
measurements, the final concentration of rhodamine 6G varied
from 20 nM to 2 µM, and the same set of samples were analyzed
with two batches of colloidal solutions. For all the IEIS samples,
SERRS and SERS Spectra of R6G-d
SERRS (top) and SERS spectra (bottom)) of both R6G-d
are shown in Figure 2. Spectra (a) and (c) pertain to 2 and 200
nM R6G-d solution, respectively, while (b) and (d) pertain to 2
and 200 nM R6G-d solution. The insets in both plots are the
expanded views of the 585-630-cm spectral region, which more
clearly shows the spectral difference induced by isotopic edition.
As evident in both the SERRS and SERS spectra, several Raman
0
and R6G-d
4
. The
0
and R6G-
d
4
0
the concentration ratio of R6G-d
5, 66.7/33.3, 50/50, 33.3/66.7, 25/75, or 0/100.
Measurement Procedures. If not specified, all SERS/SERRS
0
to R6G-d
4
was either 100/0, 75/
4
-1
2
measurements were performed using the following procedure:
after mixing of 2 mL of silver colloidal solution with 2 mL of high-
purity water in a 5 mL-glass vial, 200 µL of 2% NaCl solution was
added for aggregation followed by immediate addition of 300 µL
of analyte solution. The prepared sample was allowed to sit for 2
min before spectral acquisition. The integration times for all
4
bands are red-shifted in R6G-d relative to their locations in R6G-
d
0
, which is consistent with the higher mass of deuterium relative
2
1
to hydrogen. The peaks at 615 (ring in-plane bending ) and 771
cm (C-H bending ) of R6G-d
respectively, in R6G-d . Moreover, the singlet peak at 1356 cm
-
1
21
-1
0
are shifted to 604 and 763 cm ,
-
1
4
Analytical Chemistry, Vol. 77, No. 11, June 1, 2005 3565

Figure 3. SERRS spectra of mixtures of R6G-d0 and R6G-d4 with
a total R6G concentration of 200 nM. The concentration ratios of R6G-
d0 to R6G-d4 are (a) 100/0, (b) 75/25, (c) 66.7/33.3, (d) 50/50, (e)
33.3/66.7, (f) 25/75, and (g) 0/100. The spectra are scaled and offset
for better visualization.
Figure 2. SERRS (top, 2 nM) and SERS (bottom, 200 nM) spectra
of R6G-d0 (solid lines a and c) and R6G-d4 (dashed lines b and d).
-
1
The insets in both plots show expanded views the 585-630-cm
spectral region. Spectra are scaled and offset for clarity.
2
1
(
0
aromatic C-C stretch ) of R6G-d splits into two peaks at 1325
-
1
and 1350 cm in R6G-d
4
.
Quantitative Analysis with ISIE. Figure 3 shows the SERRS
-
1
spectra (in the 585-630-cm region) obtained with a total R6G
concentration of 200 nM but different ratios of R6G-d and R6G-
. Thus, there is a clear trend in the spectral shape and the R6G-
to R6G-d concentration ratio.
To demonstrate IEIS can also be used for SERS quantification,
0
d
d
4
0
4
similar experiment procedures were applied for SERS spectral
acquisition and analysis. Figure 4 shows the SERS spectra
obtained from samples with same R6G-d /R6G-d ratio of 50/50,
4 0
but different total R6G concentrations were (a) 20 nM, (b) 200
nM, and (c) 2 µM. Clearly, the relative spectroscopic contribution
of R6G-d and R6G-d to the mixture spectra depends only on their
0 4
concentration ratio, over a wide total concentration range. Similar
results were obtained from SERRS spectra with the same R6G-
0 4
d /R6G-d ratio with total concentrations of 200 pM and 200 nM
(data not shown).
Figure 4. SERS spectra obtained from solutions each of which has
50/50 R6G-d0/R6G-d4 concentration ratio but different total R6G
For spectral decomposition with SERRS spectra were per-
concentrations: (a) 20 nM, (b) 200 nM, and (c) 2 µM (the spectra
are offset for clarity).
-
1
formed using only the 545-674-cm spectral region. The linear
baseline of each truncated spectrum was automatically determined
from a fit the first and last three data points in the above spectral
window, and the resulting baseline-subtracted data matrix is
represented as D. The truncated and baseline subtracted spectral
component SERRS spectra. After intensity calibration (as de-
scribed in the Experiment Section), the relative concentration of
matrix S containing the pure R6G-d
0
and R6G-d
4
spectra was
R6G-d
0 4
and R6G-d in the matrix C was readily determined using
obtained in the same way from the corresponding single-
the following least-squares spectral decomposition where super-
3566 Analytical Chemistry, Vol. 77, No. 11, June 1, 2005

Table 2: Normalized SERRS Peak Intensity at 615 cm-
for R6G-d0 at Different Concentrations
1
silver
concentration of R6G-d0
colloidal
batch
200 pM
2 nM
20 nM
200 nM
1
2
3
1.00
2.16
1.27
6.11
4.70
5.83
41.32
35.76
19.86
651.55
1428.11
540.06
results were also obtained using SERS (rather than SERRS)
measurements with a total R6G concentrations of 20 nM, 200 nM,
and 2 µM (and separate batches of colloidal solution used for
calibration at a 20 nM R6G concentration and testing at the three
different concentrations).
An average error of 2.1% in the concentration ratio obtained
when the IEIS and the analyte were of the same concentration.
When all the SERRS and SERS data were considered for samples
in which the concentration difference between R6G-d
0
and R6G-
d
4
is less than or equal to 3, the average concentration ratio
prediction errors were 2.8%. Furthermore, the robustness of the
IEIS method is clearly demonstrated with the high-reproducibility
data shown in Figure 5, obtained from different batches of colloidal
solution of a 1000-fold analyte total concentration range. Moreover,
by combining SERS and SERRS measurements, the concentration
0
of R6G-d can be accurately quantified over a concentration range
of 4 orders of magnitude from 200 pM to 2 µM.
However, when the concentration difference of analyte and its
IEIS is greater than a factor of ∼3, somewhat larger concentration
ratio prediction errors were obtained. Thus, to ensure an accurate
quantification, it is important to pre-estimate the concentration of
the analyte and to make sure similar amounts of IEIS and analyte
were mixed prior to the SERS/SERRS acquisition. Note that this
can readily be done by pretesting the analyte solution of interest
using several IEIS solutions each differing by a factor of 10 in
IEIS concentration.
Figure 5. Predicted concentration ratio of R6G-d0 obtained from
IEIS SERRS measurements with the first (a) and second (b) batch
of colloidal solutions. The pure SERRS calibration spectra of R6G-
d0 and R6G-d4 were acquired with the third batch of colloidal solution.
The solid line represents the exact correlation with a slope of 1. The
error bar represents one standard deviation.
script t and -1 represents matrix transpose and inverse, respec-
tively.
As a further measure of the improvement in quantization
obtained using the IEIS method, we attempted to correlate the
t
t -1
absolute intensity of the SERRS signal of the analyte of R6G-d
with its concentration (without using an IEIS reference). The
0
C ) D‚S ‚(SS )
(4)
-1
normalized SERRS signal intensities at 615 cm obtained at each
concentration with different batches of silver colloidal solution are
shown at Table 2. In the normalization step, the difference in the
integration time for different samples was compensated and the
peak intensity for the SERRS spectrum obtained with the first
However, as described in the introduction, the SERRS intensity
depends not only on concentration but also on the characteristics
of the colloidal solution. To properly test the IEIS method, the
component spectra in matrix S were acquired with one batch of
colloid solution, and the SERRS spectra used for prediction were
obtained with different batches of colloidal solution. Furthermore,
samples of total R6G concentrations of 200 nM and 200 pM were
used to further test the robustness of the IEIR method. Figure 5
0
batch of colloidal solution and with 200 pM R6G-d was set to 1.00.
Evidently, linearity of the SERRS data obtained with any one batch
of the colloidal solution was quite poor as was the reproducibility
of the data obtained with different batches. The relative prediction
error obtained from such correlations was as large as 300% (as
determined by attempting to predict the concentration using
SERRS data for a one batch of colloidal solution that was predicted
using a intensity/concentration correlation derived from data
acquired using another batch of colloidal solution).
We have also investigated the feasibility of in situ addition of
the IEIS into the analyte solution that has been preequilibrated
with a colloidal solution. This procedure proved to be less reliable
than first premixing the analyte and IEIS and then adding the
colloidal solution. However, the results revealed interesting
0
shows the predicted percentage of R6G-d with SERRS spectra
obtained from mixtures with (a) the first batch and (b) the second
batch of colloidal solutions (and the pure component spectra in
matrix S were acquired with the third batch of colloidal solution
with a R6G concentration of 200 nM). The average and standard
deviations of each data point in both plots were obtained from 10
SERRS measurements, 5 with total R6G concentration of 200 nM
and another 5 with a total R6G concentration of 200 pM. Similar
results were obtained when the pure component spectra in matrix
S were acquired with any batch of the Ag colloidal solution. Similar
Analytical Chemistry, Vol. 77, No. 11, June 1, 2005 3567

Figure 6. SERS spectra taken when R6G-d4 was added into a
premixed R6G-d0/Ag colloidal solution (with the same R6G-d0 and
R6G-d4 concentrations). Spectra a-d were obtained at 0, 1, 80, and
2
90 min after adding the R6G-d4. Spectrum e was acquired from a
Figure 7. SERS spectra obtained with (a) pure R6G-d0, (b) pure
adenine, (c) 10 µM adenine and 100 nM R6G-d0, (d) 1 µM adenine
and 10 nM R6G-d0, and (e) 100 nM adenine and 1 nM R6G-d0. The
solution in which R6G-d0 and R6G-d4 were premixed before adding
the Ag colloid solution.
-1
615-cm peak intensities were adjusted to the same value in spectra
dynamics associated with the competition of the analyte and IEIS
for SERS/SERRS active sites of the colloidal particles. More
specifically, these tests were performed by first premixing an
analyte solution with a SERS active colloidal solution, to produce
c-e to better visualize the relative intensity differences of the adenine
and R6G features (although all three spectra have the same adenine/
R6G concentration ratio of 100/1).
a final R6G-d
preequilibration, an equal amount of the IEIS (R6G-d
into above mixture and a series of SERS spectra were acquired
at different times after the IEIS was added. The resulting evolution
of the Raman feature in the 585-630-cm spectral window is
0
concentration of 100 nM. After the 3 min of
) was added
Although the mechanism for the observed variations are not
known, these results clearly indicate that using different molecule
as a SERS internal standard is not nearly as effective as using an
IEIS derivative of the analyte of interest.
4
-
1
shown in Figure 6, along with spectrum e obtained from a sample
in which an equal amount of R6G-d and R6G-d were premixed
0 4
CONCLUSIONS
A new IEIS method for quantitative SERS/SERRS concentra-
tion measurements is presented. This method effectively corrects
for SERS/SERRS intensity variations caused by (a) poor reproduc-
ibility of the SERS substrates, (b) the complex structure and
dynamic of the SERS substrates, (c) the distance dependence of
the SERS/SERRS intensity on the analyte concentration, and (d)
variations in the collection or excitation efficiency of the Raman
system, as demonstrated with R6G and one IEIS derivative of R6G.
By combining SERS and SERRS, a concentration ratio prediction
error of less than 3% was obtained over 4 orders of magnitude of
total concentration with up to a factor of 3 concentration ratio
range. This method is shown to be reliable, reproducible, and
more sensitive than methods based on absolute SERS/SERRS
intensity correlations, with no internal standard, or using a
different molecule (rather than an IEIS) as a SERS/SERRS internal
standard.
before adding the colloidal solution. Thus, these spectra reveal
that when the IEIS is added to a preequilibrated analyte/colloid
mixture, significant time is required for the final mixture to
equilibrate, and even after 4 h, the intensity of the IEIS has not
yet reached equilibrium. Clearly such a procedure cannot readily
be used to quantify the analyte concentration, but it could be used
to study the kinetics associated with the analyte colloid binding
and exchange reactions.
As yet another demonstration of the advantages of the IEIS
method, we performed similar experiments using adenine rather
than R6G-d
measurements. More specifically, we premixed a series of solu-
tions each with a 1:100 ratio of R6G-d to adenine (because adenine
has about a 100 times weaker SERS signal). The R6G-d and
adenine were premixed before adding the colloidal solution. SERS
measurements were performed with final R6G-d concentrations
of 10 nM, 100 nM, and 1 µM. Figure 7 shows the resulting SERS
spectra obtained with (a) pure R6G-d , (b) pure adenine, (c) 10
µM adenine and 100 nM R6G-d , (d) 1 µM adenine and 10 nM
R6G-d , and (e) 100 nM adenine and 1 nM R6G-d . In contrast to
4 0
as an internal SERS standard for R6G-d concentration
0
0
0
Although the IEIS method may find many different types of
applications, we believe that it will prove to be particularly valuable
for the detection gene expression patterns and for comparative
proteomics. In particular, such application would employ isotopi-
cally edited SERS active molecules as labeling reagents for
quantitative comparisons of the amounts (concentrations) of
0
0
0
0
the results shown in Figure 4, where the relative SERS contribu-
tion of the analyte and its internal standard depends only on their
23,24
various biomolecules in samples derived from sources.
Current
relative concentration, the relative SERS contribution of R6G-d
0
and adenine varied significantly at different concentrations (even
though all the solutions had the same concentration ratio).
(
23) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R.
Nat. Biotechnol. 1999, 17, 994-999.
3568 Analytical Chemistry, Vol. 77, No. 11, June 1, 2005

comparative proteomics or gene expression studies are carried
out predominantly using either fluorescence or mass (e.g., isotope-
coded affinity tag (ICAT)) detection methods.25,26 These methods
suffer from limited dynamic range and relatively poor quantitative
capability. For example, with ICAT, ratios changing from 0.8 to
provide a life-saving early warning for environmental health
problems. Thus, there is a clear need for developing a more
accurate, robust, and sensitive technique for comparative genom-
ics and proteomics studies. The high sensitivity of the SERS
technique combining with unparallel quantitative accuracy of the
IEIS method makes SERS-IEIS a very promising biomedical
diagnostics and research technology. Such applications are actively
being pursued in our laboratories and will be reported in
forthcoming publications. Key advantages of this approach over
other tagging methods derive from the fact the chemical proper-
ties of the sample and reference are virtually identical, thus
minimizing quantization errors associated with differential optical
properties and tagging efficiencies or differences in substrate
binding or chromatographic retention.
1
.6 have been reported for samples with a theoretical ratio of 1,25
while the reported ratios for a fluorescence-based method were
2
6
0
.57-2.57. Such large prediction errors associated with current
methods prevent reliable identifications of protein or DNA, which
changes its expression level less than 2 times. However, in disease
diagnostics and drug screening applications, it can be important
to identify smaller changes in various biomarkers. For example,
reliable quantification of small changes may enable discovery of
new disease biomarkers that are not currently detectable but may
(
(
24) Julka, S.; Regnier, F. J. Proteome Res. 2004, 3, 350-363.
25) Shi, Y.; Xiang, R.; Crawford, J. K.; Colagelo, M.; Horvath, C.; Wilkins, A. J.
J. Proteomics Res. 2004, 3, 104-111.
Received for review February 24, 2005. Accepted April 1,
2005.
(
26) Karp N.; A.; Kreil D. P.; Lilley K. S. Proteomics 2004, 4, 1421-1432.
AC050338H
Analytical Chemistry, Vol. 77, No. 11, June 1, 2005 3569