Spectrometric Determination of the Refractive Index of Optical
Wave Guiding Materials Used in Lab-On-a-Chip Applications
ANDRES M. CARDENAS-VALENCIA,* JAY DLUTOWSKI, DAVID FRIES, and
LARRY LANGEBRAKE
University of South Florida, Marine Science College, Center for Ocean Technology, 140 Seventh Ave. S., St. Petersburg, Florida 33626
The design and optimization of light-based analytical devices often
require optical characterization of materials involved in their construc-
tion. With the aim of benefiting lab-on-a-chip applications, a transmission
spectrometric method for determining refractive indices, n, of transparent
solids is presented here. Angular dependence of the reflection coefficient
between material–air interfaces constitutes the basis of the procedure.
Firstly, the method is studied via simulation, using a theoretical algorithm
that describes the light propagation through the sample slide, to assess the
potentially attainable accuracy. Simulations also serve to specify the
angles at which measurements should be taken. Secondly, a visible light
source and an optical fiber spectrometer are used to perform measure-
ments on three commonly used materials in optical lab-on-a-chip devices.
A nonlinear regression subroutine fits experimental data to the proposed
theoretical model and is used to obtain n. Because the attainable precision
using this method of refractive index determination is dictated by the
uncertainty in the transmission measurements, the precision (with 95%
confidence) for mechanically rigid samples, namely glass and poly(methyl
methacrylate) (PMMA), is higher than those estimated for the elastomer
sample (in-house-molded poly(dimethylsiloxane) (PDMS)). At wave-
lengths with the highest signal-to-noise ratio for the spectrometer setup,
the estimated refractive indices were 1.43 6 0.05 (580 nm) for PDMS, 1.54
6 0.02 (546 nm) for glass, and 1.485 6 0.005 (656 nm) for PMMA.
Accurate refractive index estimations with an average precision equal to
0.01 refractive index units (RIU) were obtained for PMMA and glass
samples, and an average precision of 0.09 RIU for the PDMS molded slide
between 550 and 750 nm was obtained.
Index Headings: Refractive index; Optical materials; Plastics; Glasses;
Visible spectroscopy; Instrumental methods.
INTRODUCTION
The development of total analysis lab-on-a-chip systems has
experienced a fast growth during the last 30 years.1–6 In these
applications, optical sensing techniques are preferred due to
their minimally invasive nature, their virtually instantaneous
response, and the large variety of low-cost, small light sources
and detectors available. Glasses and polymeric materials have
been extensively used to fabricate integrated micro-total
analysis systems. Due to the ease of processing, which allows
for micro fabrication outside a clean-room environment,
glasses and polymers have been widely used to conform
micro-fluidic elements and/or to fabricate channels that host
optical fibers.3–6 In some cases, glasses and plastics readily
lend themselves to fabrication of optically functional elements,
enabling their incorporation into sensing devices, facilitating
the fabrication of total lab-on-a-chip devices.7–9 The wide
variety of transparent materials that can potentially be used for
creating micro-optical devices has created the need for
measurement systems that can optically characterize them.
Received 15 August 2005; accepted 9 January 2006.
* Author to whom correspondence should be sent. E-mail: cardenas@
marine.usf.edu
0003-7028/06/6003-0322$2.00/0
322 Volume 60, Number 3, 2006 Ó 2006 Society for Applied Spectroscopy
The knowledge of a material’s optical properties enables
modeling and potentially optimization of device performance.
For instance, the optical constants of a waveguide core and
cladding directly determine the numerical aperture of a wave-
guiding fiber and dictate its transparency. In our work towards
the development of integrated wave-guides for conducting
spectroscopic measurements on miniaturized systems, rather
than fiber integration,3,6 the index of refraction of optical
materials as a function of frequency is required. Certain
plastics’ refractive indices are known at specific wave-
lengths.10,12 However, the method used for processing
a material dictates the final optical properties.13 Varying the
curing time for polymerizing a resin, or applying additives and
dyes, as well as mechanical and irradiation processes, are
possibilities for inducing small changes in the polymeric
structure that can be used to tailor their optical properties and
hence their transmission characteristics.7,8,11,13
Standard techniques for refractive index determination of
solid materials include the prism coupling method, ellipsom-
etry, and spectroscopic techniques.14–19 Reported accuracies of
these techniques are 0.001 refractive index units (RIU) and
0.005 RIU for prism coupling method and ellipsometry,
respectively.17 These techniques can be elaborate (coupling
method), and sometimes expensive (especially ellipsometry). A
simple method for determining specific optical properties
resulting from the preparation of a film under certain conditions
is desirable. Because of this, and due to the importance of the
refractive index, techniques to measure this variable continue
to be reported.14–21 Most standard methods, as well as other
reports, use coherent monochromatic lasers. However, laser
sources are generally more expensive, require more laborious
alignment, and the results are limited to the radiation
wavelength provided by the laser. Infrared spectroscopy, which
determines both real and imaginary optical constants, is also
quickly becoming a standard methodology,18,19 but optical
constants in other ranges of the electromagnetic spectrum are
also needed. Photometry has been utilized to obtain visible
multi-wavelength refractive indices. Nussbaumer et al.17
reported a method based on immersion refractometry that
required an ultraviolet–visible (UV-Vis) spectrometer for
transmission measurements. The accuracy reported was better
than 0.01 RIU. Bass and Weidner15 and Bozlee et al.20 also
used visible spectroscopy. However, their methods required
a powered sample, and, as in Nussbaumer et al.,17 they required
solvents of known refractive indices. Other works that compete
with standard instruments’ accuracy,21–23 although useful for
refractive index determinations, are, however, specific either in
the type of mathematical analyses, sample use, or required
transducers. A methodology for refractive index determination
based on reflection losses measured indirectly via transmission
was reported by Jung and Rhee.24 From forty-five indepen-
APPLIED SPECTROSCOPY
dently measured angular observations of the polarized re-
flection components, refractive indices as well as sample
thickness were estimated. In a setup similar to the illustration in
Fig. 1, the magnitude of reflection loss as a function of the
incident angle, h (angle that is formed between the slide and the
beam) can be measured. We reported a pedagogical procedure
that evidences the refractive index influence on visible
transmitted light through transparent polymeric slides.25 Trans-
mission measurements at four angles and various wavelengths
were reported. Refractive indices could then be obtained from
transmission-versus-angle calibration curves. One of the
method’s advantages is that it does not require the availability
of sample-inert solvents with proper indices of refraction.
In this paper, we study the described transmission method for
determining multi-wavelength n values (in the visible range).
Firstly, simulations serve as a guide to determine the potentially
attainable accuracy and for the experimental setup design.
Secondly, transmission spectra at various angles are recorded
for materials common in the fabrication of lab-on-a-chip
devices: poly(dimethyl siloxane) (PDMS), poly(methyl meth-
acrylate) (PMMA), and glass slides. Multiple measurements at
various angles improve the statistics of the estimated refractive
index. The measured light signals and the simulation algorithm
are then used in a regression subroutine to determine the index
of refraction values of the slide. Proving the hypothesis that
measured transmitted signals are a sufficiently strong function
of the angle h, which allows for distinguishing the material
refractive index, even when using incoherent light, and at what
degree, constitutes this work’s rationale. Reported experimental
results illustrate the accuracy and precision, and replicated
measurement tests illustrate the method’s reproducibility.
PRINCIPLES AND MATHEMATICAL
FORMULATION
Figure 1 illustrates the transmission setup used to obtain the
refractive indices. As the collimated beam passes through the
slide, it is attenuated due to absorption and reflection. The
angle h and the sample’s complex refractive index, m (or its
components, the refractive and absorption indices, n and k,
respectively), determine the magnitude of this attenuation:
FIG. 1. Schematic of the used setup for determining refractive indices of
transparent solid slides.
m ¼ n þ ik ð1Þ
If the slide is transparent to electromagnetic radiation at a given
wavelength k, then the absorption is equal to zero (k(k) ¼ 0)
and reflection constitutes the only cause for the beam
attenuation.
Reflection coefficients depend not only on the value of the
refractive index of the sample, n1, but also on that of the
surrounding fluid (air), n0, and are quantified via Fresnel
formulas.26 The parallel, R1p, and perpendicular, Rls, Fresnel
component formulae are given by Eq. 2:
0

2 1=2 1

ðn1 =n0 Þ2 cosðhÞ þ ðn1 =n0 Þ2
sinðhÞ
B C
Rlp ¼ @

2 1=2 A
2 2
ðn1 =n0 Þ cosðhÞ þ ðn1 =n0 Þ
sinðhÞ
and
0

2 1=2 1
cosðhÞ
ðn1 =n0 Þ2
sinðhÞ
B C
Rls ¼ @

2 1=2 A ð2Þ
2
cosðhÞ þ ðn1 =n0 Þ
sinðhÞ
For unpolarized light, the total reflection loss coefficient, Rl, is
given by
Rl ¼ 1=2ðR2lp þ R2ls Þ ð3Þ
Strictly speaking, the light transmitted power, Tp, of a beam
through the slide is composed (when total internal reflection
does not occur) of various contributions that take place due to
multiple reflections. In Jung and Rhee’s work, all the
transmitted light contributions were captured by the detector
and were considered in the modeling to determine n.24 If only
the ‘‘first-order’’ transmitted beam is detected, then
Tp ¼ ð1
Rl Þ2 ð4Þ
The exponent arises from the fact that the beam encounters two
interfaces as it passes through the slide. Since our source and
detector are aligned (as shown in Fig. 1), the detector captures
mainly the first-order transmitted light. Assuming, that the first-
order Fresnel loss is the only significant attenuation of the
incident beam, the transmitted Tp can be written as
Tp ¼ Cð1
Rl Þ2 ð5Þ
C is a constant that accounts for the beam’s refraction as it
passes through the slide, which can be calculated as shown in
the Appendix.
MATERIALS, EQUIPMENT, AND
MEASUREMENT PROTOCOL
Microscope glass slides (7.5 3 2.5 cm2, bought from Fisher
Scientific Co.), PMMA (acquired in 12 3 12 in.2 plates from
United States Plastic Corp. and cut into pieces of approximately
2.0 3 5.0 cm2), and PDMS prepared slides are used for the
measurements. PDMS slides were prepared by mixing silicone
curing agent with silicone elastomer base in a 1:10 ratio
(Sylgard 184, Dow Corning). Uncured resin was poured into
plastic molds, covered to prevent contamination, and allowed to
cure overnight, which also permits air bubbles to settle out. The
resulting thin sheets are cut into samples (1 3 1 cm2). A PMMA
slide frame (with an aperture in the center) mounts the PDMS
APPLIED SPECTROSCOPY 323
flexible samples for taking the measurements. All slides are
thoroughly dusted before the measurements.
A USB 2000 fiber-optic spectrometer, two 400 lm core
diameter fiber-optic cables, a CUV-Var variable path-length
cuvette holder, and an LS1 tungsten lamp (Ocean Optics,
Dunedin, FL) were part of the measurement system. The
mounting components (NCR, Fountain Valley, CA, except as
noted) were: Mounting base, BA2; Filter holder, FH2; 1-in.
post, TR1; Turntable, RSX1; Right angle bracket, AB90; and
a 1.5 diameter mounting post and a mounting post clamp (Thor
Labs, Newton, NJ). The test fixture is assembled as described
previously.25 A 1/16-in. aperture drilled into opaque plastic
was placed in the light path to prevent spectrometer saturation.
Optical bench leveling ensures that the beam passes through
the sample center directly over the rotation axis.
In fiber-optic spectrometers, the spectral transmission, often
called transmittance, is given by27
Isample ðkÞ
Idark ðkÞ
T ¼ 100 ð6Þ
Ireference ðkÞ
Idark ðkÞ
where each I value represents the measured intensity at a given
wavelength. The dark, Idark(k) (taken with the room lighting
off, and while no light is delivered to the collecting optical
fiber), and reference, Ireference(k), spectra (transmission mea-
sured in air) correct for ambient illumination conditions,
allowing meaningful comparison between data sets. The
sample measurement, Isample(k), is taken while the slide is
positioned in the beam path. At least five spectral transmission
replicates were taken at each angle.
RESULTS AND DISCUSSION
Simulation Results. Equations 1–4 and A1–A4, in the
Appendix, completely define the transmission of light in
transparent samples as shown in Fig. 2. The feasibility of using
this set of nonlinear equations for estimation of refractive
indices is first evaluated via a set of simulated transmission
spectra as case studies. Randomly generated white noise was
added to simulated transmission spectra, and the refractive
FIG. 2. Flow chart with the strategy followed for studying the spectrometric
method presented.
324 Volume 60, Number 3, 2006
index, n, was determined by least squares best fitting
techniques. Figure 2 shows a flow diagram of the strategy
followed in this paper. The numerical implementation of the
proposed modeling has two uses: (1) it performs a theoretical
analysis on factors that could potentially affect an experiment,
and (2) it provides a first-principles interpretation algorithm for
the measurements performed. Figure 3 shows simulated
transmission curves as a function of the incidence angle and
refractive index. Figure 3a is a direct theoretical simulation
with the set of equations 1–4 and A1–A4. In Figs. 3b and 3c,
a normal distribution of randomly generated numbers, with
standard deviations of 0.01 and 0.02, respectively, and zero
mean for both cases (‘‘white noise’’), are added to the simulated
transmission to generate an ‘‘experimental-like’’ transmission.
The magnitude of the added uncertainty is within the typical
attainable precision for current charge-couple device (CCD)
fiber-optic spectrometers.28 Note that the number of angles for
which the transmission was experimentally simulated is
different for Fig. 3b and Fig. 3c, using seven and seventy
equally spaced incident angle data points, respectively. The
original set of data (Fig. 3) and the two ‘‘experimentally
simulated’’ sets of values were input into an optimization
search using the algorithm (Eqs. 1–3 and A1–A3) as the
interpretation model and n as the regressed parameter. The
regression was performed using a bounded simplex method in
a least squares sense. The search was limited to refractive index
values between 1 and 4. The regressed values are compared to
the real values in Fig. 3d. The correspondence between these
values indicates that the software numerical double precision
does not add rounding errors to the estimation. The discrete
points are regressed values obtained for the two cases with
added error. Figure 3d shows that the regressed values for the
two considered cases appeared with a similar deviation from
the actual value. Two estimators that can be used to quantify
this deviation are
" # 
Mean sum of  X N 
 
¼ ðnregressed
nactual Þ =N  ð7Þ
absolute residuals  i¼1 
and " #
X
N
2
Mean RSSQ ¼ ðnregressed
nactual Þ =N ð8Þ
i¼1
Table I has been prepared using Eqs. 7 and 8 on the data
plotted in Fig. 3d.
Effects of Positioning and Number of Angular Measure-
ments. It has been shown via simulation (using the theoretical
route in Fig. 2) that it is possible to vary both the number of
angular experimental data and the detector precision to obtain
a similar precision (estimated via the RSSQ) in the refractive
index estimation. Increasing the number of measured angular
inputs, h, increases the correlation of the data, and may lead to
the conclusion that this effectively reduces the effects of other
experimental errors. Simulation results are used to test this
contention. Randomly chosen sequences of error with a normal
distribution and different standard deviation were added to
simulated transmission curves with different numbers of
equally spaced angles. The data was then regressed, and an
average residual sum was calculated. The mean absolute
residual sum (Eq. 7) was plotted as function of the number of
angles in Fig. 4, clearly showing the effect of the number of
angles on the determined accuracy of n.
FIG. 3. (a) Simulated transmission as a function of refractive indices and incidence angle. (b) Experiment-like simulated transmission spectra; white noise added has
a standard deviation equal to 0.01. (c) Simulated experiment-like transmission spectra, white noise added has a standard deviation equal to 0.02. (d) Regressed
refractive indices as a function of the actual input value in the simulation.

The effect of the relative position of the angular measure-

ments on the accuracy of the index calculation is also
considered. It has been shown (Fig. 3) that the reflection
coefficient value, Rl, is a strong function of the input angle,
especially for the range 408 , h , 908, and is nearly constant
within the incidence angles of 08 , h , 408. Mean sum of
absolute residuals for randomly uniformly distributed errors
with different standard deviation for two different incident-
angular distributions were calculated. Figure 5 reveals that the
attainable precision in refractive index is nearly insensitive to
the incidence angle positions for the studied cases.
Effects of Imprecision in the Estimation of the Correc-
tion Factor C in Equation 5. The constant C (defined in the
Appendix) depends on both the lens radius and the sample

TABLE I. Comparison between estimated performances for determin-

ing refractive indices using a different number of angularly positioned
detectors and different white noise signals.

Case Figure 3b Figure 3c

Number of angles 7 70
Mean of the added error distribution .01 .02
Mean absolute sum of residuals 8.1 3 10
3 7.9 3 10
3 FIG. 4. Calculated mean sum of absolute residuals of determined refractive
RSSQ 1.2 3 10
4 1.1 3 10
4 indices as a function of the number of angular measurements for various
precisions in the transmission measurement.

APPLIED SPECTROSCOPY 325


FIG. 5. Mean sum of absolute residuals of determined refractive indices as
a function of the number of the precision on the transmission measurement for
two angular distributions. Data for equally spaced angles, between 08 and 908,
are labeled with open circles. The data points illustrated as triangles encompass
five equally spaced angles (between 408 and 908) and one additional
measurement at h ¼ 08.
thickness as well as on the refractive index of the sample. The
sample thickness can be obtained with inexpensive calipers that
can resolve up to 25 lm (1 thousandth of an inch). The sample
thickness measurement uncertainty influences the refractive
index determination as shown in Fig. 6a. Another aspect to
consider is the radius diameter imprecision. To provide an idea
of the effect of measuring a lens diameter value with an
imprecision of 625 lm, Fig. 6b is presented. It is expected that
variation in the actual transmission signal due to small
variations in this variable (and even small misalignments28)
will be taken into account by the fact that a reference
measurement (denominator in Eq. 6) is taken.
EXPERIMENTAL RESULTS
Figures 7 through 9 depict experimental spectra for glass,
PMMA, and PDMS, respectively, taken at eight different
incident angles. Equal angular spacing was used for the
FIG. 6. (a) Calculated mean sum of refractive index residuals for an uncertainty of 25 lm as a function of the determined refractive index for various sample
thicknesses. (b) Calculated mean sum of refractive index residuals as a function of n for the same uncertainty in measuring the diameter of various projecting lenses
in a setup similar to that shown in Fig. 1.
326 Volume 60, Number 3, 2006
FIG. 7. Transmission spectra as a function of wavelength taken at various
incidence angles for the glass slide sample.
measurements. Points below 350 nm are not shown due to the
high noise content of that data. Error bars are shown at
arbitrary points and are two standard deviations in length.
These error bars constitute an estimate of the imprecision due
to the lack of repeatability of the spectral measurements. Notice
that in general the estimated error bars are smaller for data
obtained within the wavelength ranges of 550 nm and 700 nm.
The average values of these estimates calculated for all
obtained data points as a function of wavelength within the
mentioned range have been calculated from the transmission
curve taken at each angle (Table II). Notice that the magnitude
of the error bars is similar for the glass and PMMA slides
(0.5% transmisttance), but they are, however, considerably
higher in the PDMS sample (around 2.0% transmission). The
larger imprecision in the spectral data is probably due to the
fact that the PDMS samples are molded in-house and that
PDMS is a flexible material. To obtain the transmission
measurements, the elastomer was mounted on a frame, and
even though care is taken to place the sample as flat as possible,
it might be difficult to obtain a high degree of parallelism
between the sample’s faces.
FIG. 8. Transmission spectra as a function of wavelength taken at various
incidence angles for the PMMA slide sample.
Figure 10 shows calculated refractive indices from the
transmission data for glass, PMMA, and PDMS over the
wavelength range of interest. The error bars presented in Fig.
10 constitute an estimate of experimental uncertainty of the
refractive index determination. This estimate is calculated as
Tmax;min ðkÞ ¼ Tave ðkÞ62rTexp ðkÞ
where Tmax,min represent the cases in which 2rTexp is added or
subtracted to the average transmission spectra, Tave, to
represent data with a 95% confidence if experimental errors
are normally distributed. The transmission standard deviation,
rTexp, is calculated from the experimental replicates. By
inputting the values Tmax and Tmin in the regression algorithm,
nmax(k) and nmin(k) are estimated. Since it was determined that
no numerical error (up to the fourth decimal place) was
generated from the numerical regression algorithm, the values
nmax(k) and nmin(k) constitute an estimate of the attainable
precision, which considers the experimental reproducibility of
the observed light signals.
FIG. 9. Transmission spectra as a function of wavelength taken at various
incidence angles for the PDMS sample.
compared with estimated refractive indices in Table III. Values
taken from Ref. 10 were reported with a precision of 1 3 10
4
RIU and are rounded to show two or three decimal places and
are assumed accurate in order to calculate residuals for the
ð9Þ estimated refractive index values obtained. Table III shows that
the determined refractive index values were considerably
higher in magnitude than those previously reported at the
extremes of wavelength-span readings. At these extremes, the
signal-to-noise ratio is lower in the fiber-optic spectrometer
setup, as suggested by the raw intensity data versus wavelength
in Fig. 11. The lowest residual values (0.004–0.005) occur
between 656 nm and 700 nm. The mean residual value for
refractive index values reported in between the wavelengths of
550 nm and 752 nm is less than 0.01 RIU (6.6% absolute
average deviation), but it has been reported with two decimal
places, since the average precision in that frequency range
equals 0.01 RIU. Figure 11 suggests that by using a light

jErrorbarsðkÞj ¼ nmax ðkÞ
nmin ðkÞ ð10Þ

Estimated values of the index of refraction are in agreement

with published results, proving the hypothesis that Eq. 5 can be
used as an interpretation algorithm of the described measure-
ments. The value estimated for PDMS, 1.43 6 0.05 at 580 nm,
agrees with published values.12 The regressed value of 1.54 6
0.02 for the soda lime commercial glass slide differs by 0.02
RIU from the reported value known at 546.07 nm.29 Only
information on the refractive index of PMMA as a function of
wavelength was found.10 These literature reported values are

TABLE II. Average uncertainty of transmission data (%) obtained from

five replicated measurements of each angular spectrum for the three
samples studied.

Angle, degrees 0 40 45 50 55 60 65 75 Average

Glass 0.2 0.5 0.5 0.5 0.5 0.5 0.5 0.4 0.5
PMMA 0.6 0.5 0.4 0.4 0.4 0.4 0.5 0.6 0.5
PDMS 0.8 2.0 1.0 1.0 1.0 2.0 2.0 3.0 2.0 FIG. 10. Refractive indices as function of wavelength for the materials
considered here. See the text for a description of the error-bar calculation.

APPLIED SPECTROSCOPY 327

FIG. 11. (a) Residual values and absolute deviation for refractive index
estimation as a function of wavelength obtained for PMMA. (b) Intensity
spectra showing the relative counts as a function of wavelength to illustrate the
relative magnitude of signal-to-noise ratio with the light source/spectrometer
utilized herein.
source and a spectrometer system that provides better light
detection limits and wider wavelength range, the refractive
index determination can be further enhanced. Another error
source could be compositional differences between commercial
PMMA relative to that used in the reference.10 The precision of
the method depends on the wavelength (as reproducibility in
spectrometers normally does). Calculated average precisions
(between 550 and 750 nm) were 0.02 for the glass, 0.01 for
PMMA, and 0.09 for PDMS. The higher imprecision found for
PDMS is attributed to the fact that the samples were molded,
and even though care was taken in the casting and mounting
process, the results were influenced by the flexible nature of the
material. The previously discussed uncertainty in the trans-
mission spectra for the PDMS sample suggests that the
reported method is demanding with respect to the planarity
and parallelism of the faces of the measured samples.
CONCLUSION
A relatively simple spectrometric technique for determining
refractive indices as a function of wavelength via the described
interpretation model and low-cost optical components was
presented. Uncertainty in the transmission spectral measure-
ments limits the attainable precision of the measured refractive
indices with the presented method. Average refractive index
values estimated from eight angular measurements and
comparison with previously reported values10 suggest accura-
cies down to 0.005 RIU (for PMMA). Even though other
methods have been reported to provide better precision and
accuracy than the presented methodology, the described
approach allows one to report accurate refractive indices with
a precision of at least 0.01 RIU, attractive for various
TABLE III. Comparison between refractive index values for PMMA, obtained with the method described herein, and those reported previously by
Sultanova et al.10
Wavelength nm 436 486 546
Estimated herein 1.71 6 0.08 1.66 6 0.06 1.57 6 0.03 1.50 6 0.02
Previously reported 1.50 1.50 1.49 1.49
328 Volume 60, Number 3, 2006
FIG. 12. Schematic illustrating the beam refraction as it passes through the
sample slide, evidencing the need for a correction factor due to refraction of the
beam.
applications. The method can only be used in materials that
can be formed into slide shapes with parallel faces that can be
held in the beam’s path. As this is an intrinsic limitation of the
method, it is also worth mentioning that most optical materials
used for lab-on-a-chip development are routinely molded, cast,
or cut into thin plates, making the method ideal for such
applications.
ACKNOWLEDGMENTS
Dr. Cardenas acknowledges financial support from Consejo Nac. de Ciencia
y Technologia, Univ. of South Florida (through Dr. Garcia-Rubio), and Univ.
of Guadalajara for the completion of his doctoral studies.
1. K. Huikko, R. Kostiainen, and T. Kotiaho, Eur. J. Pharm. Sci. 20, 149
(2003).
2. S. Metz, R. Holzer, and P. Renaud, Lab on a Chip 1, 29 (2001).
3. M. Janowiak, A. M. Cardenas-Valencia, M. Hall, and D. P. Fries, Meas.
Sci. Technol. 16, 729 (2005).
4. D. C. Duffy, J. C. MacDonald, J. A. Schueller, and G. M. Whitesides,
Anal. Chem. 70, 4974 (1998).
5. P. K. Dasgupta, I. Y. Eom, K. J. Morris, and J. Li, Anal. Chim. Acta 500,
337 (2003).
6. A. M. Cardenas-Valencia, D. P. Fries, X. Ding, H. Broadbent, and L.
Langebrake, 2004 Proceedings of the Micro-total Analysis Systems
Conference (Sweden, 2004), vol. 1, p. 114.
7. R. Horváth, L. R. Lindvold, and N. B. Larsen, Micromech. Microeng. 13,
419 (2003).
8. X. M. Zhao, S. P. Smith, M. Prentiss, and G. M. Whitesides, Conference
Proceedings - Lasers & Electro-Optics Society Annual Meeting-CLEO,
vol. 11 (IEEE, Piscataway, NJ, 1997).
9. K. Hosokawa, K. Hanada, and R. Maeda, J. Micromech. Microeng. 12, 1
(2002).
10. N. G. Sultanova, I. D. Nikolov, and C. D. Ivanov, Opt. Quantum Electron.
35, 21 (2003).
11. K. Hoshino and I. Shimoyama, J. Micromech. Microeng. 13, 149 (2003).
12. J. Brandrup and E. Immergut, Eds., Polymer Handbook (Wiley
Interscience, New York, 1989).
588 633 656 703 752
1.48 6 0.01 1.485 6 0.005 1.49 6 0.02 1.50 6 0.03
1.49 1.489 1.49 1.49
13. B. Schnyder, T. Lippert, R. Kötz, A. Wokaun, V. M. Graubner, and O.
Nuyken, Surf. Sci. 532, 1067 (2003).
14. I. L. Morris and T. E. Jenkins, J. Phys. E: Sci. Instrum. 22, 27 (1989).
15. J. D. Bass and D. J. Weidner, Rev. Sci. Instrum. 55, 1569 (1984).
16. A. R. Reisinger, H. B. Morris, and K. L. Lawley, Opt. Eng. 20, 111 (1981).
17. R. J. Nussbaumer, M. Halter, T. Tervoort, W. R. Caseri, and P. Smith, J.
Mater. Sci. 40, 575 (2005).
18. C.-P. E. Varsamis, Appl. Spectrosc. 56, 1107 (2002).
19. D. S. Moore, S. D. McGrane, and D. J. Funk, Appl. Spectrosc. 58, 491
(2004).
20. B. J. Bozlee, G. J. Exharos, A. E. Jimenez, and S. L. van Swam, J. Chem.
Educ. 79, 619 (2002).
21. Y. Ding, Z. Q. Cao, and Q. S. Shen, Opt. Quantum Electron. 36, 489
(2004).
22. L. A. Ageev, V. K. Miloslavskii, O. V. Tyutyunnik, and Kh. I. El’ashkhab,
Zh. Prikl. Spektrosk. 68, 270 (2001).
23. Y. R. Wang, X. M. Qu, L. Z. Cai, and B. M. Ma, J. Mod. Opt. 46, 1369
(1999).
24. C. Jung and B. K. Rhee, Appl. Opt. 41, 3861 (2002).
25. J. Dlutowski, A. M. Cardenas-Valencia, D. P. Fries, and L. Langebrake, J.
Chem. Educ., paper in press (2005).
26. G. R. Fowles, Introduction to Modern Optics (Dover Publications, New
York, 1975), 2nd ed., pp. 43, 44.
27. Operating Manual and User’s Guide: S2000 Fiber Optic Spectrometers
and Accessories (Ocean Optics, Inc., Dunedin, FL, 2000).
28. A. M. Cardenas-Valencia, R. H. Byrne, and E. T. Steimle, Sens. Actuators,
B, paper in press (2005).
29. E. Hauck, Fisher Brand Scientific, private communication (October 27,
2004).
30. F. W. Sears and M. W. Zemansky, Fisica Universitaria, Pearson Education
de Mexico S.A. de C. V. (1999).
APPENDIX
The light beam refracts upon entering the slide. The refraction
angle h 0 is related to h (Fig. 1) by
h i
h 0 ¼ sin
1 ðnair =nÞsinðhÞ ðA1Þ
The beam refraction reduces the amount of light received by
the optical detector. It can be proven that the center-to-center
distances between the detector and beam projection is30
dcc ¼ t sinðh
h 0 Þ=cosðh 0 Þ ðA2Þ
The cross-sectional area overlapping the beam projection and
the collection lens is twice the segment between the line that
crosses the intersections of the circles (shaded intersection in
Fig. 12). One of the segment’s areas can be written as
a function of the center-to-center distance and the radius of the
collecting lenses, r, in front of the detector:
h i qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Aseg ¼ r a cos ðdcc=2Þ=r
ðdcc=2Þ r 2
ðdcc=2Þ2 ðA3Þ
2
The correction factor is then given by the intersected area and
the original area of the beam reference when no sample is in
between the delivering fiber and the detector.
C ¼ 2Aseg =pr 2 ðA4Þ
APPLIED SPECTROSCOPY 329