See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/51654163

Detection of Drugs of Abuse in Saliva by Surface-Enhanced Raman Spectroscopy

(SERS)

Article  in  Applied Spectroscopy · September 2011

DOI: 10.1366/11-06310 · Source: PubMed

CITATIONS READS
73 1,367

5 authors, including:

Frank E Inscore Chetan Shende

FEI-IC Real-Time Analyzers
53 PUBLICATIONS   1,142 CITATIONS    53 PUBLICATIONS   785 CITATIONS   

SEE PROFILE SEE PROFILE

Atanu Sengupta Stuart Farquharson

Dr. Reddy's Laboratories Real-Time Analyzers
21 PUBLICATIONS   818 CITATIONS    126 PUBLICATIONS   1,560 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

UK,NIH,NASA View project

DOD I II, EPA 1I II IIB View project

All content following this page was uploaded by Chetan Shende on 16 January 2014.

The user has requested enhancement of the downloaded file.

Detection of Drugs of Abuse in Saliva by Surface-Enhanced
Raman Spectroscopy (SERS)
FRANK INSCORE, CHETAN SHENDE, ATANU SENGUPTA, HERMES HUANG, and
STUART FARQUHARSON*
Real-Time Analyzers, Inc., 362 Industrial Park Rd #8, Middletown, Connecticut 06457
Eighty drugs of abuse and metabolites were successfully measured by
surface-enhanced Raman spectroscopy (SERS) using gold- and silver-
doped sol-gels immobilized in glass capillaries. A method was developed
that provided consistent detection of 50 ppb cocaine in saliva in a focused
study. This general method was successfully applied to the detection of a
number of additional drugs in saliva, such as amphetamine, diazepam,
and methadone.
Index Headings: Drugs of abuse; Saliva; Surface-enhanced Raman
spectroscopy; SERS.
INTRODUCTION
During the past decade there has been a considerable
increase in the abuse of illicit and prescription drugs, the latter
enhanced by internet sales.1 One of the greatest dangers of drug
use is combining it with driving. More than 11% of drivers in a
2007 National Highway Traffic Safety Administration’s
(NHTSA) roadside survey tested positive for illicit drugs,2
while 18% of drivers killed in accidents tested positive for
illicit, prescription, or over-the-counter drugs according to a
2009 NHTSA survey.3 Consequently, there is a need for a
noninvasive roadside drug testing device, similar to the
breathalyzers used by law enforcement officials to measure
and estimate the blood alcohol concentration levels of impaired
drivers.4 Ideally, such a device will be able to correctly identify
the primary drugs of concern and their metabolites at relevant
concentrations (the current accepted cut-off threshold for
detection of most drugs in saliva falls within the 10–50 ppb
(10–50 ng/mL) range as defined by the U.S. Substance Abuse
and Mental Health Services Administration or SAMHSA)5
without false positives or negatives, in just a few minutes. It
should also be portable and easy to use. The last requirement
can be best met using saliva (oral fluid) as the sample medium.
This is a reasonable approach because drugs are represented in
saliva at concentrations similar to blood plasma,6,7 saliva is
99.5% water, making it easy to chemically analyze,8 and
simple saliva collectors are available.  including stimulants, antidepressants, opioids, and hallucino-
Currently, police must take a person suspected of driving
under the influence of drugs to a police station, draw a blood
sample, and perform analysis using standard laboratory
equipment, such as gas chromatography coupled with mass
spectroscopy. In many countries such as the United Kingdom,
a medical practitioner must determine whether the driver’s
condition is likely due to drugs and a test is warranted. In the
United States if a police officer suspects a person of driving
Received 28 March 2011; accepted 31 May 2011.
* Author to whom correspondence should be sent. E-mail: stu@RTA.biz.
DOI: 10.1366/11-06310
 
For example, Medimpex United Inc. (Bensalem, PA)
0003-7028/11/6509-1004$2.00/0
1004 Volume 65, Number 9, 2011 Ó 2011 Society for Applied Spectroscopy
under the influence of drugs, a Drug Recognition Expert (DRE)
may be called to the scene to profile behavior to decide if the
person warrants clinical tests (blood or urine) to determine drug
levels. Besides the legal implications, these processes are time
consuming and expensive. The availability of a rapid drug-
screening device that uses saliva as the sample would simplify
analysis and potentially eliminate the need for the medical
practitioner or DRE and unnecessary tests.
Toward developing such a device, we have been investigat-
ing the potential of surface-enhanced Raman spectroscopy
(SERS) to both identify and quantify drugs and their
metabolites in saliva.9,10 The expected success of this approach
is based on the extreme sensitivity of SERS,11,12 the ability to
measure very small samples (e.g., 0.1 mL), and the ability to
identify molecular structures of drugs through the rich
vibrational information provided by Raman spectroscopy.13
The potential for measuring and differentiating the parent drug
and its major metabolites would also provide a method of
estimating how long a drug has been in the body. In an effort to
produce a SERS-active sampling system, we have developed
and patented a method to incorporate silver and gold
nanoparticles in a porous glass structure immobilized in glass
capillaries.14,15 In contrast to most SERS substrates, the sample
does not have to be dried to achieve enhancement16 but can be
drawn into the capillary and immediately measured. Further-
more, electronegative gold and electropositive silver can be
used to attract positively and negatively charged chemical
groups, respectively, to ensure interaction of the drugs with the
plasmon field, which is critical to enhancement. While a
number of researchers have employed liquid chromatography
for chemical separations and SERS for detection in a laboratory
setting,17–19 we believe that the use of solid-phase extraction
(SPE) will allow separating the drugs from saliva in a field
environment.20 With this in mind we have been investigating
the ability of this combination with the goal of developing a
roadside drug-screening device using an SPE-based sampling
system, a SERS-active substrate, and a portable Raman
analyzer. Here we present the detection of several drug classes,
gens, extracted from saliva, by SERS, with a focus on cocaine.
EXPERIMENTAL
All drugs listed in Table Ià were obtained from Cerilliant
(Austin, TX) as 1 mg/mL methanol or acetonitrile forensic
samples and diluted in water for measurements. The chemicals
and solvents used to prepare the sol-gels were obtained at their
purest commercially available grade from Sigma-Aldrich
(Milwaukee, WI) and used as received. The SPE material
was obtained from United Chemical Technology (Bristol, PA).
à
Part of RTA’s SERS Library.
APPLIED SPECTROSCOPY
TABLE I. List of 80 relevant drugs and metabolites measured by SERS.

Stimulants Hallucinogens Benzodiazepines Antidepressants

Cocaine (HCl)c PCPc Alprazolamc Amitriptylinec
Cocaine (base)c LSDc Clonazepamc Anesthetics
Benzoylecgoninec Mescalinec Diazepamc Etomidateb
Ecgonine methyl esterb Psilocinc Flurazepamc Ketaminec
Ecgonine HClc Opiates/Opioids Flunitrazepamc GHBa
Ecgonine ethyl esterb Morphineb Lorazepamc Dextromethorphanc
Amphetaminec Heroinb Midazolamc Sedative/Pain Relief
MDAc 6-MAMc Nordiazepamc Promethazinec
Methamphetamineb Oxymorphonec Oxazepamc Scopolamineb
MDMAb Hydromorphoneb Temazepamc Diphenhydramineb
MDEAc Oxycodoneb Triazolamc Acetaminophenc
Ephedrine HClb Hydrocodoneb Estazolamb AMGa
Caffeinec Codeineb Barbiturates Naproxenec
Nicotinec Norcodeinec Amobarbitalc Ibuprofenc
Methylphenidateb Buprenorphineb Barbituric acida Acetylsalicylic acida
Phenterminec Methadoneb Barbitala Salicylic acida
Additives Meperidineb Butabarbitala Cannabinoids
Lidocaineb Fentanylc Butalbitala D9-THCb
Benzocainec Hypnotic Sedatives Hexobarbitala 11-OH-D9-THCb
Procainec Methaqualoneb Pentobarbitala 11-nor-D9-THC-CAb
Mannitolc Antipsychotics Phenobarbitalc Cannabinolb
Quininec Clonazipinec Secobarbitala Cannabidiolb
a
Active on silver.
b
Active on gold.
c
Active on both silver and gold.
Italics indicate a metabolite; nordiazepam, oxazepam, and temazepam are psychoactive metabolites of diazepam. MDA ¼ 3,4-methylenedioxyamphetamine,
MDMA ¼ 3,4-methylenedioxymethamphetamine, MDEA ¼ 3,4-methylenedioxyethylamphetamine, PCP ¼ 1-(1-phenylcyclohexyl) piperidine, LSD ¼ lysergic acid
diethylamide, 6-MAM ¼ 6-acetylmorphine, GHB ¼ c-4-hydroxybutyric acid, AMG ¼ acetaminophen glucuronide, THC ¼ tetrahydrocannabinol, 11-nor-D9-THC-
CA ¼ 11-nor-D9-THC-9-COOH.

The gold and silver sol-gels were prepared according to
previously published procedures14,15 by mixing a metal–ligand
precursor (e.g., silver amine or gold chloride) with a silicon
alkoxide precursor (octadecyltrimethoxy-silane, tetramethyl
orthosilicate, or methyltrimethoxysilane (MTMS)) in methanol.
The SERS capillaries were prepared by drawing 20 lL of the
gold- or silver-doped sol-gels into 10 cm long, 1 mm diameter
glass capillaries to produce ;1 cm plugs. The plugs were
allowed to gel and cure, after which the incorporated silver or
gold ions were reduced with dilute NaBH4.
The SPE packed capillary columns (10 cm long, 1 mm
diameter) were prepared by first drawing ;5 lL of MTMS into
one end of a glass capillary by syringe to form a porous sol-gel frit
as a support for the SPE material. Second, ;100 lL of 20 mg/mL
SPE/ethanol slurry was drawn into the capillary to form an
approximately 5 cm plug. And third, a second 5 lL of MTMS was
drawn into the capillary to secure the SPE material. The SPE
capillary column was preconditioned by sequentially flowing 1
mL each of methanol, water, and 10 mM acetic acid. Each
experiment was performed by drawing the test sample through the
SPE capillary column, washing the column to remove any
interferents, and, just prior to the final elution step using 2%
ammonium hydroxide in acetonitrile, the SERS capillary was
attached to the SPE capillary column so that the extracted drugs
could be eluted directly onto the SERS capillary.
Saliva used in preparing artificial samples was collected
from consenting employees at Real-Time Analyzers, Inc.
(RTA, Middletown, CT), using oral swabs obtained from
Medimpex United Inc. (Bensalem, PA). The swabs consisted
of a foam head attached to a syringe plunger. To collect saliva,
the foam head was placed into the person’s mouth and gently
moved around for approximately one minute to let sufficient
saliva collect in the foam. The swab was then placed into the
plastic syringe tube, and pressed to expel the saliva into a vial.
Approximately 1 mL of saliva was collected in 1 minute by this
process. The pH was measured at 6.95 to 7.05, but the exact
composition was not determined.
Drug-doped saliva samples were prepared in plastic
centrifuge tubes by spiking a small amount of aqueous drug
at the required concentration into saliva and gently vortexing
the sample for 10 seconds for uniform mixing. For example, 10
lL of 1 ppm aqueous cocaine was added to 0.990 mL of saliva
to yield a 1 mL 10 ppb cocaine doped saliva sample. The
samples were allowed to equilibrate for 30 minutes at room
temperature. Samples were measured within 60 minutes of
saliva collection.
Surface-enhanced Raman spectra for all of the drugs were
measured using a Fourier transform Raman spectrometer
(RTA, model RamanPro) that provided 100 to 3350 cm1
spectral coverage with constant 8 cm1 resolution to provide a
basic, searchable spectral library. The SERS-active capillaries
were fixed horizontally to an XY positioning stage (Conix
Research, Springfield, OR) mounted above a fiber-optic probe.
The probe delivered 75 mW of 785 nm laser excitation to the
capillary and collected the 1808 scattered radiation. Raman
spectra of cocaine residue were also measured using this
spectrometer with 300 mW laser power and a 5 min acquisition
time. Further instrument details have been published.13 To
demonstrate field measurement capability, cocaine was also
measured using a 5 lb battery-operated, hand-held dispersive
Raman analyzer (RTA, model SERS-ID) that provided 517–
1753 cm1 spectral coverage with variable resolution from 1.03
to 1.55 cm1. The SERS capillary was fixed horizontally in the
sample compartment of the portable analyzer, which delivered
45 mW of 785 nm excitation laser to the capillary and collected
the 1808 scattered radiation. Figure 1 compares the SERS of
cocaine measured using the two different Raman instruments.

APPLIED SPECTROSCOPY 1005

FIG. 1. SERS of 10 ppm cocaine measured on (A) hand-held Raman analyzer
(RTA, model SERS-ID) and (B) Fourier transform Raman spectrometer (RTA,
model RamanPro). Conditions: (A) 45 mW at 785 nm (50 lm diameter spot), 5
s acquisition, 1.03–1.55 cm1 resolution and (B) 75 mW at 785 nm (250 lm
diameter spot), 1 min acquisition, 8 cm1 resolution.
RESULTS AND DISCUSSION
Eighty drugs of abuse and metabolites were measured using
both silver- and gold-doped sol-gel capillaries. Although many
drugs were highly active on silver, such as phenobarbital, c-
FIG. 2. SERS of (A) phenobarbital, (B) GHB, (C) LSD, (D) mescaline, (E)
diazepam, (F) MDA, (G) PCP, and (H) heroin. Conditions: 1 ppm in water,
(A–D) in silver-doped sol-gel capillaries, (E–H) in gold-doped sol-gel
capillaries; 75 mW at 785 nm (250 lm diameter spot), 1 min acquisition, 8
cm1 resolution.
1006 Volume 65, Number 9, 2011
FIG. 3. Comparison of SERS for (A) 100 ppb cocaine (as cocaineHCl) in
gold-doped sol-gel capillary and (B) 50 ppm cocaine in silver-doped sol-gel
capillary to (C) RS of pure cocaine. Conditions: (A and B) as in Fig. 2; (C) 300
mW of 785 nm, 5 min acquisition, 8 cm1 resolution. Spectra were intensity
normalized for this stack plot by setting the baseline to 0 and the 999 cm1 peak
height to 1.
hydroxybutyric acid (GHB, date rape drug), lysergic acid
diethylamide (LSD), and 3,4,5-trimethoxyphenethylamine (mes-
caline), the majority of the drugs were active on gold, such as
diazepam (Valium), 3,4-methylenedioxyamphetamine (MDA), 1-
(1-phenylcyclohexyl) piperidine (PCP), and heroin (Fig. 2).
Cocaine was active on both metals, but considerably more so on
gold (Fig. 3). It was easily measured at 100 ppb (;6 pg in the 60
nL scattering volume) on gold, but only at 50 ppm on silver.
Enhancement factors were calculated to be 3.1 3 106 and 1.1 3
104, respectively, by comparing these spectra to the normal Raman
spectrum (RS) of pure cocaine and taking into account the relative
999 cm1 peak intensities (baseline corrected), laser power, and
concentrations. Consequently, gold-doped sol-gel capillaries were
chosen for further method development.
In an effort to determine the idealized sensitivity of the analysis,
cocaine samples diluted in water to 10 ppb were measured using
five capillaries at each concentration. It was found that consistent
spectra were obtained at 50 ppb and higher (Fig. 4), whereas
measurements at 10 ppb produced sporadic results. Furthermore,
the spectral intensities (baseline-corrected peak heights), could be
FIG. 4. SERS of cocaine on gold-doped sol-gel filled capillaries at 100, 75, 50,
25, and 0 ppb (top to bottom). All spectra are averages of 5 capillary
measurements. Spectral conditions: as in Fig. 2.

FIG. 5. Plot of 999 cm1 peak height as a function of cocaine concentration
(A) in saliva (SPE treated, green squares) and (B) in water (red diamonds,
corresponding to spectra in Fig. 4). The average standard deviation for all
concentrations was (A) approximately 625 ppb for cocaine in saliva and (B)
615 ppb for cocaine in water.
fit with a straight line below 100 ppb (Fig. 5) but followed a
Langmuir–Blodgett curve at higher concentrations (i.e., rolled-
over as monolayer coverage of the surface was approached).9 The
average standard deviation for all of the measurements was
approximately 615 ppb.
Next, a series of saliva samples doped with cocaine
beginning at 1000 ppm, then diluted by factors of 10, was
measured. Again, five gold-doped sol-gel capillaries were used
for each concentration. The lowest concentration that gave
consistent measurements was 10 ppm. This was expected, since
salivary mucins (a class of high molecular weight glycosylated
proteins) can trap drug molecules due to their highly viscous
nature and clog the sol-gel pores, as well as block the metal
surface. Furthermore, certain basic drugs can also bind to alpha
acid glycoproteins in saliva,21,22 thus making the bound drug
molecules unavailable to the metal surface.
Previously, several methods were investigated to break up
mucins and release drugs from proteins, and it was found that
the combination of acetic or sulfuric acid and heat or sonication
was effective.10 However, towards the goal of developing an
easy-to-perform field method, alternative pretreatment methods
were examined, specifically, solid-phase extraction. SPE
capillary columns were prepared using 1 mm glass capillaries
to match, and be used in-line with, the SERS-active capillaries.
An SPE material consisting of silica particles functionalized
with octyl groups and benzyl sulfonic acid groups was chosen
because it has been successfully used to separate numerous
drugs including cocaine from biological fluids.23
Again, a series of saliva samples doped with cocaine was
prepared, pretreated using SPE capillary columns, and
measured (five times at each concentration). For this method,
the following steps were preformed. (1) A 1 mL saliva sample
(500 lL of saliva plus 500 lL of 10 mM acetic acid) was
drawn through a 1 micrometer filter into a SPE capillary
column, depositing the cocaine and passing some of the saliva
components. (2) 1 mL each of 10 mM acetic acid, methanol,
and water were drawn through the SPE capillary column
removing any residual saliva components. (3) A 20 lL 2%
ammonium hydroxide in acetonitrile solution was drawn
through the SPE capillary column into a SERS capillary,
extracting the cocaine from the SPE, and depositing it on the
FIG. 6. SERS of (A) 50 and (B) 0 ppb cocaine in saliva after using SPE
pretreatment. Spectral conditions: as in Fig. 2.
gold-doped sol-gel. (4) The SERS capillary was placed in the
sample compartment of a Raman analyzer and measured. The
entire process took less than 10 minutes for each sample. Using
this procedure, cocaine could again be consistently measured at
50 ppb and higher and sporadically at 10 ppb. However, this
time the 999 cm1 cocaine peak heights were consistently
higher by approximately a factor of 2 than those obtained for
water (Figs. 5 and 6), suggesting that the SPE material not only
separated the cocaine from the mucins and alpha acid
glycoproteins but also concentrated it.
To test this idea, cocaine samples in water were also
measured using the above SPE method. It was found that the
signal intensities were approximately four times those obtained
without the SPE. This indicates that not all of the cocaine is
successfully extracted from saliva. Furthermore, it was found
that saliva samples devoid of cocaine produced a fairly flat
background, but that the baseline-corrected peak height at 999
cm1 was not always zero, presumably due to baseline noise
(Fig. 6). Nevertheless, cocaine could be detected at the required
concentration of 50 ppb. In fact, this is best illustrated by a
receiver-operator characteristic (ROC) curve (Fig. 7, often used
by government agencies24). A ROC curve is a plot of the
specificity versus sensitivity, which is expressed as the
probability for a binary system of false positive detection (x-
FIG. 7. Receiver-operator characteristic curves for cocaine in saliva at (A) 75,
(B) 50, and (C) 25 ppb. The dashed line represents a 50/50 chance of true and
false positive detection.
APPLIED SPECTROSCOPY 1007

FIG. 8. SERS of (A) amphetamine, (B) diazepam, (C) methadone, and (D)
PCP. Conditions: 1 ppm in saliva on gold-doped sol-gel capillaries; 75 mW at
785 nm (250 lm diameter spot), 1 min acquisition, 8 cm1 resolution.
axis) versus the probability of true positive detection (y-axis),
respectively. Here, the sensitivity is defined as the threshold at
which the cocaine SERS peaks can be positively detected
above the background noise 90% of the time.4 ROC curves
were prepared from the multiple measurements of cocaine in
saliva at each concentration previously described. As shown in
Fig. 7, the probabilities of detection for 75 ppb, 50 ppb, and 25
ppb with 90% confidence are 100%, 92%, and 66%,
respectively. This general method was successfully used to
detect 1 ppm amphetamine, diazepam, methadone, and PCP,
all artificially added to saliva as shown in Fig. 8. As part of this
research, a software program was also developed to search and
match the measured SERS spectra as a library. The spectral
library also included a number of potential interferents, such as
aspirin and caffeine. Nevertheless, it was found that the
software, using the Euclidean distance algorithm,25 easily
identified the target drugs with sufficient discrimination.
CONCLUSION
A method was successfully developed that allowed consis-
tent detection of five different drugs in saliva at 1 ppm or less
within 10 minutes. In a focused study, ROC curves were used
to establish that cocaine could be measured at 50 ppb more
than 90% of the time. This concentration is 5000 times more
sensitive than previously reported SERS measurements of
cocaine in saliva.10 The method employed a solid-phase
extraction capillary column to separate the drugs from the
saliva prior to delivery into a SERS-active capillary. Future
work will incorporate the capillaries into a lab-on-a-chip as part
of a sample kit to be used by police in conjunction with a hand-
held SERS analyzer for roadside testing.
1008 Volume 65, Number 9, 2011
View publication stats
ACKNOWLEDGMENTS
The authors are grateful for the support of the National Science Foundation
(DMI-0215819), National Institutes of Health (1R43CA94457-01), the UK
Home Office Scientific Development Branch, and Drs. Helen Turner and Sarah
Lamping for insightful discussions.
1. National Institute of Drug Abuse, Research Report Series, ‘‘Prescription
Drugs: Abuse and Addiction’’, http://www.nida.nih.gov/researchreports/
prescription/prescription6.html (accessed 22 March 2011).
2. R. Compton and A. Berning, ‘‘Results of the 2007 National Roadside
Survey of Alcohol and Drug Use by Drivers’’, National Highway Traffic
Safety Administration’s National Center for Statistics and Analysis, Report
Number DOT HS 811 175 (2009).
3. National Highway Traffic Safety Administration, ‘‘Drug Involvement of
Fatally Injured Drivers’’, NHTSA’s National Center for Statistics and
Analysis, Report Number DOT HS 811 415 (2010).
4. N. Isles, ‘‘A Guide to Type-Approval Procedures for Drug Screening
Devices used for Transport Law Enforcement in Great Britain’’, Home
Office and Forensic Science Service, VA4.2 (2007).
5. MD Rockville, ‘‘Substance Abuse and Mental Health Services Adminis-
tration’’, Results from the 2009 National Survey on Drug Use and Health:
Volume I. Summary of National Findings (Office of Applied Studies,
NSDUH Series H-38A, HHS Publication No. SMA 10-4586Findings),
available at: http://www.oas.samhsa.gov (2010).
6. O. Drummer, Forensic Sci. Int. 150, 133 (2005).
7. D. Crouch, J. Day, J. Baudys, and A. Fatah, ‘‘Evaluation of Saliva/Oral
Fluid as an Alternate Drug Testing Specimen’’, National Institute of
Justice, USA, Report Number 605-03 (2004).
8. J. Chamberlain, The Analysis of Drugs in Biological Fluids (CRC Press,
Boca Raton, FL, 1995), 2nd ed., p. 46.
9. S. Farquharson, C. Shende, F. Inscore, P. Maksymiuk, and A. Gift, J.
Raman Spectrosc. 36, 208 (2005).
10. C. Shende, F. Inscore, A. Gift, P. Maksymiuk, and S. Farquharson, SPIE
6007, 165 (2005).
11. S. Nie and S. Emory, Science 275, 1102 (1997).
12. K. Kneipp, Y. Wang, H. Kneipp, L. Perelman, I. Itzkan, R. Dasari, and M.
Feld, Phys. Rev. Lett. 49, 1667 (1997).
13. S. Farquharson, A. Gift, C. Shende, F. Inscore, B. Ordway, C.
Farquharson, and J. Murran, Molecules 13, 2608 (2008).
14. S. Farquharson, Y. H. Lee, and C. Nelson, U.S. Patent 6,623,977 (2003).
15. S. Farquharson, A. D. Gift, F. E. Inscore, and C. S. Shende, U.S. Patent
7,393,691 (2008).
16. H. Turner and S. Lamping, ‘‘Commercial SERS Substrates for a Roadside
Screening Device for Drugs in Oral Fluids’’, Proceedings of the Joint
Meeting of the International Council on Alcohol, Drugs and Traffic Safety
and The International Association of Forensic Toxicologists, Seattle
(2007).
17. R. Freeman, R. Hanmaker, C. Meloan, and W. Fateley, Appl. Spectrosc.
42, 456 (1988).
18. L. Cabalin, A. Ruperez, and J. Laserna, Anal. Chim. Acta 318, 203 (1996).
19. R. Sheng, F. Ni, and T. M. Cotton, Anal. Chem. 63, 437 (1991).
20. C. Shende, F. Inscore, A. Sengupta, J. Stuart, and S. Farquharson, Sens.
Instrum. Food Qual. 4, 101 (2010).
21. R. Kaliszan, A. Nasal, and M. Turowski, Biomed. Chromatogr. 9, 211
(1995).
22. J. Kala, W. Pang, H. Zubaidah, and O. Hashim, Int. J. Mol. Sci. 11, 4488
(2010).
23. N. Badawi, K. Simonsen, A. Steentoft, I. Bernhoft, and K. Linnet, Clin.
Chem. 55, 2004 (2009).
24. W. Liu, S. Wang, C.-I. Chang, J. L. Jensen, J. O. Jensen, H. Knapp, R.
Daniel, and R. Yin, SPIE 5995, A.1 (2005).
25. P. Gemperline, Practical Guide to Chemometrics (Taylor & Francis
Group, CRC Press, Boca Raton, FL, 2006), 2nd ed.