|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
DEVELOPMENT, OPTIMIZATION AND CHARACTERIZATION OF
A SURFACE ENHANCED RAMAN SPECTROSCOPIC METHOD
FOR DETECTION OF DIPICOLINIC ACID
JOY D. GUINGAB
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Joy D. Guingab
To my parents--my \roenlgth and foundation
I would like to express my sincere gratitude to my research adviser Prof. James
D.Winefordner for his continued guidance and encouragement. I am especially grateful to have
had the opportunity to work with a mentor who encourages personal growth through research
I would like to acknowledge Dr. Nicolo Omenetto for his valuable research guidance.
His admirable love for science is truly inspiring. I am very grateful to Dr. Benjamin Smith for
all his advice in building my experimental set up. He has set a very good example for all the
students in the Winefordner group.
I would like to extend my appreciation to my committee members for their valuable
advice on the completion of this research dissertation.
The Winefordner-Omenetto group is worthy of thanks for the camaraderie and friendship.
I am especially grateful to Dr. Jamshid Temirov, Benoit Lauly and Nicholas Taylor for all
Raman discussions and the assistance extended during the setting up of the Raman system. I am
thankful that I have had the opportunity to be in the same group as Dr.Xihong Wu for the past 3
years and have developed a great friendship. Many thanks go to Dr. Kirby Amponsah-Manager,
Ron Whiddon and Pam Monterola for their constant encouragement and friendship.
I would like to thank Dr. David Hahn and Dr. Timothy Anderson for allowing me to use
their Raman systems in the preliminary stages of this research. I am grateful to Young Seok Kim
for all the assistance in the use of their Raman system. All the UV-Vis measurements were
conducted in laboratory of Dr. Nigel Richard, Dr. Charles Martin and Dr. Kathryn Williams.
Thanks go to the SEM laboratory staff for all training and assistance in the use of the SEM
I would like to extend my appreciation to all support staff in chemistry, including the
glass, machine and electronics shop for all the work their contributions in the instrument
development. It has been great having Ms. Lori Clark all these years as the most helpful person
to every Chemistry graduate student. Special thanks go to Ms. Jeanne Karably for all her
assistance and great advice on almost anything.
My life in Gainesville had not been this memorable without the company of good friends.
I acknowledge all my friends in chemistry for all the good times shared. I am so fortunate to
have a very good support group from my fellow Filipino graduate students and their families.
My heartfelt gratitude goes to the Javelosas for being so accommodating, to Jemy and Fair for
their friendship and constant encouragement, to Jhoana and Dodge for their good company and
editorial and computer-related assistance, and to John for his thoughtfulness. A special
appreciation is owed to my dear friend Rina for being the best roommate and more importantly
my best friend. Those years of friendship will always be treasured and cherished.
I am grateful to my relatives for all their support. I am very grateful to the Flores family
in Montreal for all the support extended during my graduate studies and for being my family
away from home. I would like to thank the Lamar family for graciously accommodating me in
their home in Jacksonville and for treating me as a family member.
I am blessed with a family who endlessly supports and believes in me. I would like to
acknowledge my brothers, Jaymar and Jun for their love and support. My brothers and I were
surrounded by a tremendous wealth of love and support from my grandparents while growing up.
I am forever grateful to my mother for her incredible strength and unconditional love. Without
her believing in my potential, I would never have reached this point. I owe my father gratitude
for instilling in me the strength that I need to face the challenges in life. He has been my driving
force in all my accomplishments. Much appreciation is extended to my husband, Emil for all the
support and encouragement and most importantly for his unwavering friendship and love.
TABLE OF CONTENTS
ACKNOWLEDGMENT S .............. ...............4.....
LI ST OF T ABLE S ................. ...............9................
LIST OF FIGURES .............. ...............10....
1 INTRODUCTION ................. ...............15.......... ......
Back ground and Si gnifi chance ................. ................. 15......... ....
Scope of Research Dissertation ................. ...............18.......... ....
2 THEORETICAL OVERVIEW .............. ...............20....
Raman Scattering............... ...............2
Historical Background............... ...............2
Spontaneous Raman Effect ................. ...............21........... ....
Raman Signal Enhancement Techniques .............. ...............26....
Surface Enhanced Raman Spectroscopy (SERS) .............. ...............27....
Historical Back ground ............... ...............2
Enhancement Mechanisms .............. ...............27....
Substrates for SERS ................ ......... ...............3
Advantages and Limitations of SERS .............. ...............35....
3 DESIGN, OPTIMIZATION AND CHARACTERIZATION OF A RAMAN SYSTEM
FOR SERS APPLICATION................ ..............4
Introducti on .............. .. ......__ ..... ...............40
Design, Optimization and Characterization............... ............4
Excitation Source................ ......... ..........4
Sampling Mode and Collection Optics............... ...............43.
M onochromator ................. ...............44.......... ......
Optical Filters .............. ...............47....
D etector .............. ...............49....
Data Collection ................. .............. ....... ........ ..........5
Evaluation of Performance of the Raman System ................. ...............54........... ..
4 STABILITY OF SILVER COLLOID AS SUBSTRATE FOR SURFACE ENHANCED
RAMAN DETECTION OF DIPICOLINIC ACID .............. ...............83....
Introducti on ................... ......... ...............83.......
Experiments and Methods .............. ...............84....
Results and Discussion .............. ...............85....
5 EVALUATION OF EXPERIMENTAL CONDITIONS FOR THE SURFACE
ENHANCED RAMAN DETECTION OF DIPICOLINIC ACID (DPA) ON SILVER
COLLOIDS GENERATED BY FLOW INJECTION ANALYSIS (FIA) ..........................100
Introducti on ................ .. ........ ..... ...............100......
Experimental Design and Methods............... ...............102
Results and Discussion ................... ........... ...............104......
Factors Affecting SERS of DPA ............ .....__ ...............104
FIA-SERS of DPA .............. ...............109....
Conclusion ............ ..... ._ ...............109...
6 FEASIBILITY OF SERS DETECTION OF BACTERIAL SPORES .............. ...............122
Introducti on ................. ............. .. ...............122.....
Experimental Methods and Procedures .............. ...............124....
Results and Discussion ................ ...............124................
7 CONCLUSIONS AND FUTURE WORK ................. ...............133........... ...
Conclusions............... .. .............13
Future Research Directions............... ..............13
LIST OF REFERENCES ................. ...............138................
BIOGRAPHICAL SKETCH ................. ...............145......... ......
LIST OF TABLES
3-1 Common lasers used in Raman Spectroscopy. ............. ...............57.....
3-2 Nominal output power of the visible lines of the Argon ion laser ................. ................58
3-3 A summary of the important characteristics of the Spex 1680B double
monochromator. .............. ...............66....
3-4 The monochromator scan speed at different increments. ............. .....................6
3-5 A summary of the characteristics of the Hammamatsu R-928 PMT. ............. ..............73
4-1 A summary of normalized signal intensities and calculated enhancement factors from
SERS of DPA on silver colloids. ........... ..... .__ ...............99.
6-1 A survey of SERS methods for detection of dipicolinic acid and bacterial spores .........132
LIST OF FIGURES
2-1 Energy diagram illustrating the Raman effect. ............. ...............37.....
2-2 A schematic diagram of surface enhanced Raman process .............. .....................3
2-3 A simplified schematic diagram illustrating the SERS electromagnetic enhancement
m echanism .............. ...............39....
3-1 The Raman experimental set-up .............. ...............56....
3-2 The Argon ion visible lines ..........._ ..... ..__ ...............59.
3-3 The experimental set-up used for measuring Argon ion laser profile and laser short
and long term fluctuations. ............. ...............60.....
3-4 The short-term fluctuation of the 5 14.5 nm line of the Argon ion laser ..........................61
3-5 The long-term fluctuation of the 5 14.5 nm line of the Argon ion laser. ..........................62
3-6 A schematic diagram of the 900 sampling geometry adapted for Raman system ..............63
3-7 A diagram illustrating (a) the position of the dove prism in the Raman system and (b)
the image rotation by a dove ............. ...............64.....
3-8 A comparison of normal Raman spectra of saturated DPA illustrating the
improvement in signal intensity due to the image rotation using a dove prism. ...............65
3-9 Transmission spectra of different types of optical filters used in Stokes Raman
Spectroscopy ................. ...............68.......... .....
3-10 A comparison of the optical density plot of a long wave pass filter vs. a notch filter.......69
3-11 A spectrum of Ar' laser with a Iser transmitting filter ................. .......... ...............70
3-12 The decrease in the intensity of Rayleigh scatter peak using a high pass filter .................71
3-13 A comparison of the Rayleigh peak attenuation by two different long wave pass
filters. ............. ...............72.....
3-14 A typical pulsed height distribution. Disc 1 and Disc 2 are the discriminators of the
photon counter .......... ............... .............. .................. ...............74
3-15 The pulse height distribution for the photon counting system for determining the
discrimination l evel ................. ...............75......___ ...
3-16 Benzene spectra obtained using analog and photon counting mode illustrating the
difference in signal to noise. ............. ...............76.....
3-17 A timing diagram used for the photon counter and monochromator interfacing and
synchronization ........... ..... .._ ._ ...............77....
3-18 Mercury lamp spectra obtained with (a) analog and (b) photon counting mode
showing the narrow mercury lines ................. ...............78........... ...
3-19 Raman spectra of several pure solvents measured using the analog mode........................79
3-20 Raman spectra of 1.0 x 10-4 M beta-carotene in (a) THF and (b) ethanol ................... ......80
3-21 A normal Raman spectrum of saturated DPA in 1 M KOH. ............. .....................8
3 -22 Raman spectra of an empty cuvette, deionized water and 1 M KOH ........._..... ..............82
4-1 Comparison of the normal Raman and SERS spectra of dipicolinic acid (DPA).............. 91
4-2 SERS signal of DPA on silver colloids aged over 1 h .............. ...............93....
4-3 Absorption spectra of silver hydrosol before and after addition of dipicolinic acid
(D PA ). .............. ...............94....
4-4 Absorbance of silver colloids at the peak maximum 401 nm as a function of sodium
b orohy dri de. .............. ...............9 5....
4-5 SERS signal of DPA on silver colloid with and without sodium sulfite as oxygen
scavenger. ........... ..... ._ ...............96...
4-6 Absorbance of the silver colloids at the peak maximum 402 nm. ............. ...............97
4-7 SEM images of silver colloids prepared using sodium borohydride treated with
sodium sulfite. ........... ........... ...............98....
5-1 The Flow Inj section Analysis (FIA)-SERS set up ................. ...............110........... .
5-2 Absorbance readout recorded from the FIA system ................. .......... ................1 11
5-3 SERS signal of DPA at 1013 cml on silver colloid prepared at different volume
5-4 Absorption spectra of silver colloids prepared at different volume ratios of sodium
borohydride to silver nitrate ........._.__...... ..__. ...............113..
5-5 SEM images of silver colloid prepared by batch addition with different volume
ratios ................. ...............114................
5-6 SERS signal of DPA at 1013 cml on silver colloids generated at different flow rates ..1 15
5-7 Scanning electron micrographs of silver colloid prepared by batch addition and
generated by FIA system at different total flow rates ................. .......... ...............116
5-8 The effect of addition of different concentrations of nitric acid and potassium
hydroxide on the SERS signal of DPA at 1013 cm~l ................ ....__ ................1 17
5-9. The effect of aggregation of silver colloid upon addition of different concentrations of
sodium sulfate on the SERS of DPA. ................ .....__ ....................1
5-10 The effect of aggregation of silver colloid upon addition of different concentrations
of sodium chloride on the SERS of DPA. .....__.....___ ..........__ .........19
5-11 A typical FIA peak profile of DPA measured at 1013 cm-l using the FIA-SERS
system .............. ...............120....
5-12 Comparison of percent relative standard deviation of SERS of DPA under static
condition and FIA-mode ................. ...............121....... ......
6-1 The SERS spectrum of 25 ppm DPA used for estimating the limit of detection
6-2 SERS spectra of different concentrations of DPA. ......____ ........__ ................129
6-3 A plot of DPA signal (1013 cm l) vs concentration (0-78 ppm) using the unknown
peak (966 cm-0) as an internal standard. .......................... ........130
6-4 SERS spectra of 78 ppm DPA measured using freshly prepared and 2-day old silver
colloids. ........... ..... ._ ...............131...
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
DEVELOPMENT, OPTIMIZATION AND CHARACTERIZATION OF A
SURFACE ENHANCED RAMAN SPECTROSCOPIC METHOD
FOR DETECTION OF DIPICOLINIC ACID
Joy D. Guingab
Chair: James D. Winefordner
Major Department: Chemistry
Surface Enhanced Raman Spectroscopy (SERS) has been explored as a tool to study
biological agents. The anthrax distribution in the US postal in October 2001 intensified the need
for a rapid and accurate detection of bacterial spores. Dipicolinic acid (DPA) is a unique
component of bacterial spores and has been used as a signature molecule for detection schemes.
The feasibility of using SERS as a detection technique for bacterial spores is evaluated in this
A conventional Raman system was designed, optimized and characterized. The lab-
constructed Raman system was used for the SERS studies of DPA on silver colloid dispersions.
Experimental conditions required to obtain reproducible and intense SERS signal from DPA
molecule were evaluated and optimized. A continuous flow system was constructed for SERS of
DPA studied under flow mode. Improvement in SERS signal reproducibility was demonstrated
with the use of controlled conditions of a custom-built Flow Injection Analysis (FIA) system.
The developed FIA-SERS system allowed silver colloid generation, sample introduction and
DPA detection. Characterization of the silver colloids was done by absorption method and
Scanning Electron Microscopy (SEM).
The feasibility of using SERS as a detection technique for bacterial spores was evaluated.
The absolute limit of detection was estimated to be 40 ng or 106 Spores which is 2 orders of
magnitude higher than the infectious dose set by the Centers for Diseases Control and Prevention
and required limit of detection for bacterial spores. The signal enhancement factor of the
prominent DPA peak at 1013 cml was approximated to be 2 orders of magnitude increase from
the normal Raman signal of saturated DPA. Although, quantitative analysis of DPA was not
demonstrated in this research, Raman peaks attributed to DPA molecules were observed in
concentration as low as 25 ppm.
Background and Significance
Airborne microorganisms, which exist in respirable size of 1-10 Clm, can cause adverse
effects on humans and can potentially cause fatal disorders. One of these biological agents is the
bacterial spore. Bacteria are known to thrive almost everywhere. There is a wide array of
bacterial species ranging from the benign to the infectious such as Bacillus anthracis, the
causative agent for anthrax. Bacteria are microscopic but complex organisms. Aside from their
ability to multiply exponentially, they can adapt to any kind of environmental stress by triggering
some response mechanisms.l When exposed to unfavorable growth conditions such as depleting
food and water supply or extreme temperatures, physiological changes and cellular responses are
triggered to combat starvation and other harsh conditions. Failure of these immediate responses
is the onset of a period of metabolic dormancy and resistance to stress or sporulation. Bacterial
spores can stay dormant for a few hours to millions of years until conditions are favorable for
germination. Metabolic dormancy and resistance to chemical agents, radiation and heat are the
main reasons for their long-term stability.2 Endurance to all possible harsh environmental factors
makes bacterial spores a potential hazard to human health. The recent anthrax attack in 2001
shows the potential danger that spores may pose.
Substantial progress towards the development of rapid detection methods for bioaerosols,
particularly bacterial spores, has been made over the past few years. The need for rapid and
accurate detection of these airborne contaminants has been intensified by the anthrax distribution
in the US postal system in October 2001. Detection of these biological agents is extremely
important in defense against bioterrorism, prevention of disease outbreak as well as monitoring
in industrial or clinical setting. Time consuming and tedious microbiological assays based on the
morphology of the microorganism are commonly used for analysis of biological agents. Routine
identification of spores includes nucleic acid sequencing using the Polymerase Chain Reaction
(PCR) technique.3-5 Immunoassays are commonly employed as spore identification
techniques.6,7 Although these assays are nondestructive and sensitive, they require sample
preparation and do not allow automation of analysis. Fluorescence spectroscopy, which plays an
important role in the bioanalytical Hield, has also been used.s-10 Laser Induced Breakdown
Spectroscopy (LIBS) has also been explored for its feasibility to study bacterial spores.11-1
However, these methods are not capable of distinguishing bacterial spores from other
interference such as pollens and molds.
Dipicolinic acid (2, 6-pyridinedicarboxylic acid or DPA) in the form of calcium
dipicolinate (CaDPA) is known to be a unique constituent of bacterial spores.2 Techniques
geared towards the detection of CaDPA and its derivatives as Bacillus utin acltl i\ signature are
now being developed. Sensitive and highly selective detection methods, which involve
extraction of DPA from spores such as mass spectrometry.14-16 A DPA triggered terbium
luminescence has been employed as a method for rapid detection of DPA extracted from
bacterial spores.l7l The drawback of these methods is that sample preparation is time
consuming and often tedious, making the analysis time longer. The Hield of vibrational
spectroscopy has been explored as a tool for whole organism Eingerprinting; the most promising
is Raman spectroscopy.l
As a versatile, information-rich technique, Raman spectroscopy plays an important role
in the identification and characterization of microorganisms. The advent of Raman microscopy
even makes it possible to probe and characterize an individual bacterium20 as well as a bacterial
spore.21 Detection of a single bacterial spore using Raman spectroscopy coupled with optical
trapping has been investigated.22 Although Raman spectroscopy can provide rich information
about a molecule, its major disadvantage is that of having a very weak signal due to the small
cross section of the interaction process. In fact, the Raman cross section is approximately 12-14
orders of magnitude lower than the competitive effect, fluorescence.23 Not until the 1970s, when
an unexpectedly high signal enhancement was observed from pyridine on a roughened silver
electrode24, had Surface Enhanced Raman Spectroscopy (SERS) shown promise to overcome the
traditionally low sensitivity of normal Raman spectroscopy.
In the past decades, studies have been conducted to understand the signal enhancement
observed when a molecule is attached to various metals such as silver, gold and copper. The
signal enhancement has now been increased to 14 orders of magnitude making single molecule
detection possible.24-26 This signal enhancement would translate into a Raman cross section of
approximately 10-16 CM2/mOlecule. There are two operative mechanisms responsible for the
SERS phenomenon, namely electromagnetic field enhancement and chemical enhancement. The
former contributes to the enhanced Raman signal when scattering takes place in the enhanced
local optical fields of the metal and the latter is due to metal-molecule interaction. With the
inherent selectivity of Raman spectroscopy combined with signal enhancement, SERS has
become a valuable analytical tool with low detection limits and short analysis time. Moreover,
unlike with infrared (IR) measurements, water does not interfere, making SERS very useful in
Substrates used for SERS evolved from roughened silver electrodes to a variety of
substrates prepared in various ways.27 Both chemical reduction28-3 and photo reduction34-36 have
been used to generate metal hydrosols. Laser ablation of metal films has also been used for
preparation of colloidal sols.37-41 Metal vapor evaporation42,4 and surface etching44-4 have been
employed to prepare solid support-based SERS substrates. Metal colloids as SERS substrate is
of particular interest due to straightforward preparation and simplicity of characterization and
manipulation.47 However, it has been shown that extreme dependence of the SERS effect on the
physico-chemical properties of the colloids imposes a rather severe restriction on the usefulness
of SERS on colloids as an analytical technique.30 It has been shown that a continuous flow
configuration SERS can improve the reproducibility of the colloid preparation, thus improving
Bacterial spores studies using SERS has been reported recently. It has been used as a
tool for probing specific biochemical components and for discrimination of bacteria.51,52 DPA
extracted from spores53,54, as well as in whole bacterial spores has also been detected.19,55 SERS
of a single bacterial spore has been made possible by optical trapping.56
The long term goal of this research is to detect DPA in bacterial spores. In the present
study, a SERS method was developed for detection of DPA. Controlled conditions of a custom-
built flow injection system allowed precise preparation of the colloid, controlled aggregation and
thorough mixing of the colloid and the bacterial spores. This approach addressed the problem of
poor reproducibility common in silver colloid preparation that leads to non-linearity of the SERS
signal with the analyte concentration while providing a sensitive detection method for bacterial
spores. The feasibility of SERS detection of 10000 spores, which is the infectious dose for
Bacillus untinut~l i\5 was evaluated.
Scope of Research Dissertation
The focus of this dissertation is on the development, optimization and characterization of
a Surface Enhanced Raman Spectroscopy system for the detection of bacterial spores. The
research involved SERS of DPA studied both in static conditions and continuous flow mode.
Static condition SERS studies of DPA were vital in the development and optimization of the
Flow Inj section Analysis (FIA)-SERS system.
Theoretical considerations of Raman spectroscopy are important before focusing on
SERS and are addressed in Chapter 2. This includes both classical and quantum mechanical
description of Raman Spectroscopy. An overview of SERS including the enhancement
mechanisms involved is discussed in Chapter 2. A variety of SERS substrates are available but
the use of metal colloids is investigated in this dissertation and briefly covered in Chapter 2. The
design, optimization and characterization of a conventional Raman system for the SERS study
are presented in Chapter 3. In Chapter 4, the stability of silver colloids as SERS substrate for
DPA detection is evaluated. Optimization of conditions for SERS of DPA in FIA mode is
presented in Chapter 5. The capability of SERS detection of DPA in spores is evaluated in
Chapter 6. General conclusions and recommendations for future directions are described in
A brief overview of Raman scattering is necessary in order to understand the theory
behind surface enhanced Raman spectroscopy. Several Raman reference books were consulted
in writing this section of the dissertation chapter.23,58-61
In the 1920s, researchers were interested in the scattering of light by charged particles.
The Compton effect which explained the changes in the wavelength of X-ray photons when
scattered by electrons was first documented.62 There was also a related study predicting this
process by Kramers and Heisenberg in 1925.62 Theoretical studies conducted by Smekal in 1923
led to the first prediction of the Raman effect.62 Chandrasekhar Venkata (C.V) Raman in India
first experimentally observed Raman scattering in 1928.63 The investigation involved focusing
sunlight and using filters while observing color changes in the scattered light. Simultaneous to
the discovery by C.V Raman, a study about Raman scattering was reported by Landsberg and
Mandelstam in Moscow.62 C.V Raman initially referred to the scattering process as "Feeble
Fluorescence" and later termed the phenomenon "New Radiation." The importance of this
discovery was recognized by the awarding of the Nobel Prize in Physics to C.V Raman in 1930.
Ever since, it has been known as Raman Spectroscopy, Raman effect, or Raman scattering.
Following these pioneer investigations, more fundamental studies and applications of
Raman spectroscopy had been conducted. Until the mid 1980s, the Raman literature had limited
reports on the chemical applications of the technique. The limiting factors were both
fundamental and technical. Fundamentally, Raman scatter is inherently weak and can be easily
overwhelmed by fluorescence. Technical issues were addressed by the introduction of PMT-
photon counting detection systems and the laser. The feasibility of routine chemical analysis
became possible after the recent introduction of Fourier-transform (FT)-Raman, charged coupled
devices (CCD), small powerful computers and the availability of near infrared lasers.
Spontaneous Raman Effect
Raman and Brillouin scattering are two types of inelastic scattering. Raman scattering is
caused by rotational and vibrational transitions in molecules. Rotational transitions are lower in
energy and slower than vibrational transitions. The molecule may have collisions with other
molecules during rotational transition causing a change in the rotational state of the molecule.
Rotational Raman spectroscopy is carried out in gas phase at low pressure to ensure that the time
for collisions is greater than the time for a transition. Relatively large frequency shifts which are
independent of scattering angle are observed in Raman scattering. Brillouin scattering, which is
caused by thermal fluctuations of the medium, yields small frequency shifts that are dependent
on the scattering angle.
Raman spectroscopy is an inelastic light scattering effect due to the interaction of light
with a sample. When a sample is illuminated with intense monochromatic light, photons are
reflected, absorbed and scattered. It is the scattering process that gives information about the
structure of a molecule. The light can be scattered in three ways. Most of the scattered radiation
has the same energy as the incident light resulting to elastic scatter which is referred as the
Rayleigh scatter. A small fraction of these photons is scattered inelastically by either losing
energy to the molecule or gaining energy from the molecule. Photons which lose energy give
rise to Stokes Raman scattering and photons which gain energy give rise to anti-Stokes Raman
scattering. The energy gained by the molecule in Stokes Raman scattering appears as vibrational
energy. When a molecule has excess vibrational energy above the ground state, it is this energy
that is lost to the anti-Stokes Raman scattering. A Raman spectrum is a plot of the scattered light
intensity as a function of the Raman shift from the incident light in cm- The Raman shift is
calculated with respect to the frequency of the excitation light using a general equation
where vo is the wavenumber of the excitation source, v is the Raman wavenumber of the
sample, h0 is the wavelength of the laser and h is the wavelength corresponding to the Raman
data of the sample. This equation illustrates that the Raman shift is independent of the
wavelength of the excitation source. Both elastic and inelastic scatter can be observed in a
Raman spectrum and are symmetrically positioned on each side of the Rayleigh scatter.
Although the Stokes and anti-Stokes lines are symmetrically positioned about the Rayleigh peak,
their intensities are different with the Stokes being more intense. The Rayleigh peak appearing
at 0 cm-l can be greatly attenuated by a band rej section filter. The Raman effect has been
described using both classical and quantum mechanical model. The discussion of the two
models in this dissertation serves as a general overview.
When an incoming radiation with an oscillating electric field impinges on a molecule,
this induces a change in the dipole moment of the molecule. When a molecule is in the presence
of an oscillating electric field, its electron cloud is distorted resulting in an induced dipole
moment. The induced polarization pu is proportional to the strength of the electric field, E as
given in the equation
p = aE (2-2)
where the proportionality constant, a, corresponds to the polarizability of the molecule.
Polarizability is related to the ease with which the electron cloud can be distorted when a
molecule is subj ected to an electric field. The electromagnetic radiation generates a fluctuating
dipole moment of the same frequency and can be expressed as
E = E, cos mi,t = E, cos(2xivot) (2-3)
where E, is the electric field strength at equilibrium and mi, is the angular frequency of the
electromagnetic radiation. Equation 2-2 is a generalization and only describes the induced dipole
that has a vibrational frequency v, equal to the frequency of the incident light vo. This induced
dipole will scatter light at the same frequency as the laser light which is the Rayleigh scatter.
Molecular vibrations are considered to be composed of normal modes, Q1
.Q, = Qocos2xyrVt (2-4)
where vl is the characteristic harmonic frequency of the jth normal mode. For a molecule with N
atoms Q, is equal to 3N-6 and for linear molecules Q, is equal to 3N-5. During vibration, the
changes in polarizability and the amplitude of the change is given by
a ~ a + 6~2,a : +...(2-1
When a molecule interacts with the incident radiation with an electromagnetic field strength E,
equation 2-2 becomes
pu = aE = aE, cos 2xivot (2-6)
Substituting equations 2-4 and 2-5 into 2-6 yields:
p~~ = coE coxot OE, cos 2xyr,t cos 2xyv (2-7)
Using the identity
[cos(x + y) + costx y)l
cos xcosy = (2-8)
equation 2-7 can be expressed as:
pu = aoE, cos 2xivot + Eg acs@oV)cs@o- 29
This equation summarizes that the induced dipole will radiate light at three frequencies. The first
term in the equation is the Rayleigh scatter which has the same frequency as the incident
radiation and magnitude proportional to the inherent polarizability of the molecule, ao The
second term describes the anti-Stokes Raman scatter occurring at a frequency equal to
(vo + v, and the third term is the Stokes Raman scatter occurring at (vo v,). The classical
description of Raman scattering although incomplete, provides some useful insights. It indicates
a linear dependence of polarization and both Rayleigh and Raman scattering intensities on the
intensity of the incident radiation. At higher values ofEo nonlinear Raman scattering can occur.
Another insight is that only molecular vibrations that cause a change in polarizability of the
moleule ieldRama scatein s 0 This has been the basis of the primary selection
rule of Raman in contrast to IR absorption which requires a change in the dipole moment. The
signs in the equation indicate that Raman scatter can be positive or negative. The intensities of
Stokes and anti-Stokes are determined mainly by the Boltzmann distribution. Anti-Stokes
scattering is dependent on the population of the first vibrational excited state. The population of
the ground state is always higher than that of the excited state resulting in a more intense Stokes
scattering intensity. The difference in intensity of Stokes and anti-Stokes scattering is given by
the ratio predicted based on the Boltzmann distribution indicating the dependence on temperature
I,=, exp ,ih (2-10)
IR oV o, I\,-V kT
where h is the Planck' s constant, k is the Boltzmann constant and T is the temperature.
Another important insight from the classical description is that the change in polarizability
Smay vary significantly for different molecules and for different modes in a given
molecule. Lastly, Rayleigh scatter is always more intense than the Raman scatter so is
expected to be much smaller than the inherent polarizability of the molecule a"
Quantum mechanical model
The quantum mechanical treatment of Raman scattering recognizes that the vibrational
energy of a molecule is quantized according to the relationship
AE +hA v= AE+ h(vo F,) (2-11)
where AE is the net energy change of the molecule and h~v is the net energy change of the
photons. The mechanism of the Raman process can be illustrated simplest by an energy level
diagram. As depicted in Figure 2-1, when radiation interacts with a molecule, it is excited to a
virtual state. The virtual state is not a true quantum state of a molecule but can be considered as
a short lived distortion of the electron cloud caused by the light oscillating electric field.
Relaxation of the molecule from the virtual state to the ground state leads to Rayleigh scattering.
The energy of Rayleigh scatter is the same as the energy of the incident radiation. Raman
scattering is shown in which light scattered is lower (Stokes) or higher (anti-Stokes) in energy by
the amount equal to the vibrational transition hot. Stokes Raman scattering results from
molecules relaxing to an energy level higher than the original energy level and yields a positive
AE. On the other hand, anti-Stokes Raman scattering yields a negative AE since the molecule
relaxes to an energy level lower than the original. As discussed in the classical description of
Raman scattering, Stokes has higher in intensity than anti-Stokes. The energy population is
governed by the Boltzmann distribution. Molecules at room temperature are in the lowest
vibrational level in the electronic ground state. The intensity difference of Stokes and anti-
Stokes are discussed in the previous section.
Raman Signal Enhancement Techniques
Although Raman spectroscopy can provide rich information about a molecule, it is
inherently a weak process. Having a weak signal is a maj or disadvantage of the technique due to
a rather small cross section of the interaction process. In fact, the Raman cross section is
approximately 6-8 orders of magnitude lower than fluorescence.
Several processes can be used to enhance the sensitivity of Raman spectroscopy such as
resonance Raman and surface enhanced Raman spectroscopy (SERS). Resonance Raman effect
is observed when the laser' s wavelength is tuned to the absorption wavelength of the molecule.
The intensity of the vibrational modes is affected by the associated electronic transition.
Resonance Raman spectroscopy is very useful in the analysis of biological molecules as it is both
selective and sensitive. Its sensitivity is 2-6 orders of magnitude higher than normal Raman
spectroscopy. The weak Raman process has a cross-section between 10-31 and 10-29 CM2 ST-1
molecule- Moreover, this enhancement technique is limited by the availability of lasers that
can be tuned to be resonant with the electronic transitions of the molecule. Surface enhanced
Raman spectroscopy has shown promise to overcome the traditionally low sensitivity of
spontaneous Raman spectroscopy. A remarkable signal enhancement is observed when a
molecule is attached to various metals such as silver, gold and copper (Fig. 2-2). Twenty years
after the discovery of SERS, the signal enhancement factor approximated to be 3-4 orders of
magnitude improved to about 14 orders of magnitude.26 This would translate into a Raman cross
section of approximately 10-16 CM2/ ST-1 moleCUle-1. The succeeding section of this chapter of the
dissertation provides a general overview of the theory of SERS based on several theoretical
Surface Enhanced Raman Spectroscopy
An intense Raman signal from pyridine adsorbed on electrochemically roughened silver
electrodes was observed by Fleishmann, Hendra and McQuillan in 1974.24 This remarkably
strong signal was attributed to the presence of a large number of pyridine adsorbed on a greater
surface area provided by the surface of the roughened electrode. Three years later, two papers
published separately reported an increase in Raman signal from pyridine adsorbed on silver
surfaces and attributed the signal enhancement to an intrinsic enhancement mechanism. In
contrast to the conclusion made earlier, the strong Raman signal was caused by an increase in
Raman cross section in the presence of an adj acent rough metal structure and not to the increase
in number of pyridine molecules adsorbed on the surface of the electrode. These independent
investigations of Jeanmaire and Van Duyne7 and Albrecht and Creighton72 led to the emergence
of what is known as surface enhanced Raman spectroscopy (SERS).
SERS Enhancement Mechanisms
After the discovery of the SERS effect, many researchers have conducted fundamental
studies on the mechanism behind the signal enhancement. Understanding of the SERS
mechanisms has been a subject of many ongoing research projects. Electromagnetic field
enhancement and chemical enhancement are the two operative mechanisms that have been
proposed to explain the increase in the Raman cross section of molecules adsorbed on metallic
nanostr-uctures. It has been observed that some form of roughness is required for SERS.
Equation 2-9 shows that the Raman Stokes component of the polarization contains both the
polarizability and the electromagnetic field intensity. The local field intensity relates to the
electromagnetic field enhancement mechanism while the inherent polarizability of the molecule
relates to the chemical enhancement mechanism. The observed signal enhancement is
contributed by the combination of these two mechanisms. The contribution by each mechanism
depends on the molecule and the optical and chemical characteristic of the SERS substrate.
In normal Raman, the total Stokes Raman signal is proportional to the Raman cross
section as well as the laser intensity. This relationship is given by
Ph (Vs) c~ No~ree L) (2- 12)
where Pa (v,)is the total Stokes Raman signal in W sr- N is the number of molecules present
in the probe volume, G~re is the Raman cross section in cm2 ST-1 moleCUle-land I(v, ) is the
laser intensity in W. This expression can be modified to describe the total SERS Stokes
signal Pssn vs) Based on the proposed SERS mechanisms, equation 2-12 can be expressed as
Pssn (s) acx N' oR A(v, ) 2 IA(vs ) 2 ~(L ) (2-13)
where N' is the number of molecules involved in the SERS process, oR, is the increased Raman
cross section. The terms A(v, ) and A(vs ) are enhancement factors for the laser and for the
Raman scattered field, respectively.
Electromagnetic field enhancement mechanism
In understanding the electromagnetic enhancement mechanism, the size, shape and
material of the surface have to be taken into consideration since these characteristics determine
the resonant frequency of the conduction electrons in the metallic particle. When a metallic
nanoparticle is exposed to electromagnetic field of the same frequency, electrons on the surface
oscillate. Metals have many free electrons which vibrate or oscillate when excited by an
electromagnetic wave at a particular frequency. This collective resonant oscillation known as
plasmons generates an additional electric field on the surface of the metallic particle.
Electromagnetic field enhancement mechanism is based on the idea that scattering takes place
within the enhanced local optical fields of the nanoparticle. A simplified schematic diagram
(Fig. 2-3) aids in understanding the electromagnetic field enhancement mechanism.26 It shows a
spherical metallic nanoparticle with a dielectric constant, o,~ present in a medium with a
dielectric constant, e. The particle has a diameter of 2r which is smaller than the wavelength of
the incident laser. A molecule positioned at a distance, d, from the surface of the spherical
particle experiences an electric field, En This field is the superposition of the incident electric
field Eo and the induced field on the surface of the particle, EI
E, = Eo + Es (2-14)
The field induced on the metal sphere is dependent on the dielectric constants of the medium and
the particle, the size of the particle as well as the distance of the molecule to the particle. This
relationship is given by
Es = r 3E (2-15)
S+ 2eo a(r+ d)3
The field enhancement factor A(v) can be expressed as the ratio of the field at the position of the
molecule and the incident field.
EEv e+ 2E r+d (-6
Based on this equation, the conditions of the resonant excitation of the surface plasmon can be
described. Maximum field enhancement factor can be achieved when the real part of the
dielectric constant of the particle is equal to 2Eo. This can be achieved for silver and gold
particles using wavelengths within the visible and near infrared region. The imaginary part of
the dielectric constant needs to be small.
If the enhancing factors for the laser and the Stokes field enhancement are taken into
account, an expression for the electromagnetic enhancement factor for the Stokes signal can be
G,,,,(vs)= A)vL AFvs) L o S i o2 12 (2-17)
E(VL)+ 2Eo I I(s)+ 2Eo r+d
Several useful insights describing the electromagnetic enhancement mechanism can be deduced
from equation 2-17. It indicates that the enhancement scales as the 4th pOWeT Of the local field of
the metallic particle. This can be maximized when the excitation and the scattered field are in
resonance with the localized surface plasmon of the metal. Electromagnetic enhancement
mechanism does not require the molecule to be adsorbed directly on the metallic surface but the
distance dependence is extremely strong scaling as a 12th factor. An enhancement factor of 104
1012 has been attributed to the electromagnetic enhancement mechanism
Chemical enhancement mechanism
Chemical enhancement mechanism requires direct adsorption of a molecule on the
surface of the metallic nanostructure. The contribution to the total enhancement has been
estimated to be a factor of about 100. This mechanism is associated with the overlapping of the
electronic states of the metal and the adsorbed molecule. A charge transfer state is created
between the metal and the adsorbed molecule. The conduction band of the metal is divided by
the Fermi level. This is the boundary between the filled and unfilled orbitals within the
conduction band which moves up or down depending on the electric potential of the metal. An
adsorbate molecule that is widely separated from the metal has discrete energy levels. When a
molecule is adsorbed on the metal, the orbitals overlap to form a bond between the metal and the
adsorbate. This orbital interaction permits electrons to tunnel from the adsorbate molecule to the
metal. This process is referred to as charge transfer. The charge transfer state increases the
probability of a Raman transition to occur by providing resonant excitation.
The chemical enhancement mechanism via the charge transfer theory has several
characteristics that distinguish it from electromagnetic field enhancement mechanism. The
charge transfer theory indicates that the enhancement is due to the interaction of the metal and
the molecule and not the optical properties of the metallic nanostructure. A site specific
interaction between the metal and the adsorbate is required for chemical enhancement. The more
prominent characteristic is that chemical enhancement is short range in nature and limited to the
first layer of adsorbed molecules.
Since the discovery of SERS in 1970, the SERS-active substrates have evolved from the
electrochemically roughened silver electrode to a variety of forms. The pioneer works on both
electrodes and colloidal systems have provided the foundations of SERS theory. Substrates for
SERS, commonly noble metals such as silver, gold or copper are used in colloidal form or
immobilized on surface.27 For these substrates to be SERS-active, attention must be given to the
type and preparation procedure. Based on the discussion of the electromagnetic enhancement
and chemical enhancement mechanisms, factors that can affect the SERS signal include the
dielectric constants of both the metal and its medium, the morphology of the metallic surface, the
distance between the adsorbate molecule and the metal surface and the excitation frequency.
According to the electromagnetic enhancement mechanism, SERS intensity is also affected by
excitation of the local field of the metal surface. Therefore, it is critical to control the factors that
influence the excitation of the localized surface plasmon resonance (LSPR) of the metal in order
to achieve the maximum SERS signal.
Formation of metal colloids is very simple and easy; not requiring sophisticated
instrumentation. Colloidal systems are well suited for solution phase SERS measurements.
They can be easily characterized using absorption spectroscopy. There is definite dependence of
SERS signal on the size and shape of the nanoparticles. The use of metal colloids for analytical
detection has several limitations. SERS has been shown to require a certain degree of colloid
aggregation. The aggregative state of the metal can easily be affected by the introduction of the
analyte. Not only does SERS depend on the manner of introduction of the analyte but also
depend on the exact preparation procedure of the metal colloids. It is not surprising that SERS
on metal colloids have been found to be irreproducible.
Metal colloids or hydrosols can be prepared by chemical reduction of simple salts of
silver or gold. Reducing agents include borohydride28-32,48,49 and citrate.33 Turkevich et al used
citrate as a reducing agent to produce gold colloids of uniform size and spherical shape.73 By
controlling the concentration of citrate, gold colloids of different sizes were produced. Gold
colloid preparation normally involves heating and vigorous stirring while that of silver colloids is
simple. However, methods for preparation of silver particles of uniform size and spherical
shapes are less developed. Silver colloids prepared from chemical reduction of silver nitrate and
sodium borohydride are usually a mixture of irregular sizes and shapes.74 Photo-reduction has
also been used to obtain metal colloids.757 Laser ablation of thin metal films submerged in
deionized water has been an alternative way to generate metal colloids that are free of
Characterization of metal colloids normally involves their absorption (extinction) spectra.
Silver colloid solutions are yellow and have a single extinction band at 385 nm while gold
colloids are wine red with an extinction band at 520 nm.74 These extinction bands result from
the localized surface plasmon resonance (LSPR) excitation of the metallic nanoparticles and
depend on their size and shape. The extinction spectrum of a metal colloid is a contribution of
both absorption and scattering of the light incident to the particles. When visible light
illuminates a small particle, the particle can absorb and scatter the light. The absorption
contribution (Cabs) to the extinction of small particles of sizes within the Rayleigh scattering
regime (a < 20 nm) is given by
where a is the radius of the particle, E= 81 + ie2 is the dielectric constant of the particle at the
optical frequency relative to that of the surrounding medium and h is the wavelength of the light
in the medium. For a colloidal solution with N particles per unit volume, the absorption is given
A = NCabs1
where 1 is the path length. The scattering contribution ("s= ) to the extinction is expressed as
1287/a6 l_ 12
sea 321 e + 2 (2-20)
In the Rayleigh scattering regime, both absorption and scattering contribution are at their
maxima when the real part of the dielectric constant is equal to -2. Absorption predominates
over scattering when the particle size is within this regime. On the other hand, as the radius
increases (a> 20 nm), scattering is the main contributor of the extinction.
In the context of SERS using colloidal particles, the shift of the extinction band to longer
wavelength is of particular interest rather than that of an isolated particle at the Rayleigh regime.
The shift in excitation resonance can be achieved by using bigger particles or aspherical or
particles in aggregates. Theoretical calculations of the LSPR of triangular, aggregated and
irregularly shaped particles were conducted to provide insight into the electromagnetic
enhancement mechanism by nanoparticles.83
Surface based SERS substrates
Different methods have been used to prepare surface based SERS substrates. The
following are the methods that are being used by different groups:27
Vapor or vacuum deposition
Corrosive etching or electrochemical roughening
Aqueous sol deposition
Noble metal embedding
Metal colloid monolayer assembly
Colloid multilayer formation
The wide range of surface SERS substrates allows researchers to study SERS in diverse
environment. A great amount of work has been focused on the development of more stable,
novel SERS substrates that can promote optimum SERS enhancement.
Advantages and Limitations of SERS
SERS as an analytical technique has many advantages. The advantages of normal Raman
Spectroscopy also apply for SERS. As a vibrational spectroscopy technique, a SERS spectrum
provides information about the molecular structure. It addresses the traditionally poor sensitivity
of normal Raman spectroscopy. SERS requires the molecule to be adhered to or near the metal
surface thus providing non-radiative pathways for the decay of excited states and thus, quenching
fluorescence. The abrupt decay of the electromagnetic fields ensures that only the molecules
adsorbed or close to the SERS substrate are being probed making SERS a selective technique.
SERS has been demonstrated to be very useful in the analysis of biological samples which are
normally in aqueous environment since water does not give a strong SERS signal.
The limitations of SERS have to be considered as well. A maj or limitation is linked to
the poor reproducibility of SERS substrates preparation making correlation of theoretical and
experimental SERS challenging. The dynamic changes in metal colloids due to aggregation of
the particles lead to irreproducible signal intensities. Production of assembled nanoparticles still
exhibits some degree of irreproducibility. The roughness feature that is required for SERS needs
to be characterized properly in order to understand the signal enhancement. Quantitative SERS
analysis has been restricted by the poor linearity of the technique. The limited choice of active
SERS substrates that exhibit surface plasmon resonance is another major constraint. Other
substrates aside from gold, silver and copper have been explored. In order for them to be used as
SERS substrates, a thin coating of an active substrate is necessary which complicates the
preparation procedure. SERS applicability is limited to molecules that naturally adsorb
chemically or physically onto the surface of the SERS substrate. Manipulation of these
molecules must be done in order for them to be within the enhanced optical field of the substrate.
High background continuum is often observed in SERS spectra. Although the source of the high
background is still under debate, the background can overwhelm the SERS signal from the
Anti-Stokes Rayleigh Stokes Raman
Raman Scattering Scattering Scattering
Figure 2-1. Energy diagram illustrating the Raman effect.
Excited electronic state
Figure 2-2. A schematic diagram of surface enhanced Raman process showing molecules
adsorbed on the surface of an aggregate of gold particles. The particle size is between
Laser a aa
E = E + lS
Figure 2-3. A simplified schematic diagram illustrating the SERS electromagnetic enhancement
mechanism. The metal is spherical and its size is smaller than the wavelength of the
laser (adapted from reference # 26).
DESIGN, OPTIMIZATION AND CHARACTERIZATION OF A RAMAN SYSTEM FOR
High-end commercial Raman instruments which meet different set of needs are currently
available. Instrument choice depends on the intended application particularly the sample
requirement and the research obj ectives. In the effort to develop a system that can be used in a
wide range of applications, the design of commercial instruments becomes complex which
results in difficulty of operation and increased system costs. The long-term goal of this research
is to detect bacterial spores using Surface Enhanced Raman Spectroscopy (SERS). The
application requirement is a simple, conventional Raman system with the basic components to
study the SERS of bacterial spores. A Raman instrument consists of Hyve basic components: an
excitation source, a sample illuminating system, a sample holder, a wavelength selector, a
detector and a data acquisition system.84,85 Although modifications such as computer interfacing
for instrument control and data collection is common, the basic components remain unchanged in
the maj ority of applications.
A schematic of the lab-constructed Raman instrument is depicted in Figure 3-1. It
consists of a Spectra-Physics BeamLok 2060 Ar+ laser (Mountain View,CA) and a SPEX 1680B
double monochromator (Horiba JobinYvon; Edison NJ). A thermoelectrically cooled
Hammamatsu R-928 (Bridgewater, NJ) photomultiplier tube coupled with a SR 400 gated
photon counter (Stanford Research; Sunnyvale,CA) was used as a detector. A dove prism
(Edmund Optics; Barrington, NJ) tilted at a 450 angle was used to rotate the image making it
parallel to the entrance slit of the double monochromator. A LabVIEW program was used for
photon counter control and data acquisition. All measurements were done using a standard 1 cm
x 1 cm quartz cuvette. SERS spectra were recorded in 900 collection geometry. Characterization
and optimization of the Raman system are covered in the succeeding sections in this Chapter.
Since selection of components based on different criterion was not the obj ective of the research,
thorough discussion of all existing Raman instrument components is not included in this
dissertation. This chapter is dedicated specifically to describe the characteristics and the
performance of the Raman system designed for the particular SERS application.
Design, Optimization and Characterization
A variety of lasers are currently available for Raman spectroscopy. The discovery of the
laser has been one of the technological advances behind the renaissance of Raman spectroscopy
as a useful analytical technique. Table 3-1 shows the wavelengths and output power ranges of
the most common commercial lasers used in Raman Spectroscopy.86 The most common of these
lasers are the Ar~ and Kr' lasers. Their popularity in Raman spectroscopy has been attributed to
their high output power, variety of output wavelengths, and relatively long lifetime. As shown in
Table 3-1, different alternatives are available depending on the application requirement.
Spectra-Physics BeamLok 2060 Ar+ laser with a maximum multi-line output power of 7
watts was used as an excitation laser. The laser is equipped with a broadband optic that allows
output of multi lines. Single line operation is done by using a prism instead of a broad band
optic. The prism disperses the laser beam separating the lines based on their wavelengths.
Adjustment procedures must be followed to get the desired wavelength which can be confirmed
by their relative power. Maximum output power specifications of the visible lines of the Ar
laser are summarized in Table 3-2.87 Visible single line output power specified for 488 and
514.5 nm are 1.8 W and 2.4 W, respectively. The laser power that reaches the sample was
approximately half of the original power due to some losses in the beam alignment and focusing.
All spectra obtained in this research were acquired using a laser power range of 100-800 mW.
The laser profie was obtained using a fiber optic and CCD spectrometer (Ocean Optics;
Dunedin, Fl). Figure 3-2 shows the laser profile of the Argon ion laser. Specifieations of the
Argon ion laser provided a power stability of a 0.3 % and 5 % over 2 h and 8 h period,
respectively." Depending on the lifetime of the laser, the performance may change over time.
While lasers are good sources of monochromatic light, fluctuations are inherent in the output
power and relatively old lasers may be more susceptible to these fluctuations. For this reason,
short-term and long-term fluctuations for the Ar' laser were considered in system
characterization. The same experimental set up used to record the laser profile was used to
perform the measurement (Fig.3-3). For short term fluctuation measurement, the laser signal
was monitored for 1000 s. For long-term fluctuation analysis, the laser signal was measured
over 5000 s. The integration time was set at 100 ms. The percent relative standard deviation
(%RSD) was calculated for each measurement. The short-term laser fluctuation, which yielded
7% RSD, indicated that for shorter analysis times, the noise contributed by the laser was minimal
(Fig.3-4). This is important for determining the system's limiting noise source. The calculated
%RSD for the long-term fluctuation analysis was 8% (Fig.3-5). This can provide an insight
about possible laser drift during long measurement time such as the DPA SERS spectra
Sampling Mode and Collection Optics
The sampling geometry and collection optics are very important in the design of a Raman
sy stem. The alignment of the laser, the sample and collection aperture has a huge influence on
the magnitude, reproducibility and S/N of the Raman signal. Since the Raman signal is
inherently weak, it is best to have an efficient collection of the scattered photons.
Raman sampling modes are divided into three categories: conventional, remote and
microscopy modes.86 COnventional sampling involves liquid samples in cuvettes or solids such
as powders and pellets. Remote sampling involves the use of fiber optic cables which can be any
usable length for remote Raman detection within meters or kilometers away from the
spectrometer. The third mode of sampling uses a microscope. With the use of a microscope
obj ective, Raman microscopy allows probing of small regions usually less than 1 Cpm in diameter
and a few Cpm in depth. For this dissertation, the conventional mode was used because the
samples were in liquid form.
Common conventional sampling geometries are the 90o and 1800 geometries. The 900
geometry was chosen for this research. With this geometry it is easy to align the laser focal
cylinder with the monochromator entrance slit. Figure 3-6 depicts the arrangement of the optics
used. A transparent sample is placed in a cuvette and the laser is directed by a mirror on the
sample cuvette. The laser focal cylinder is imaged on the entrance slit of the monochromator
with a slit width of 3 pm. With the two lenses, the image is perpendicular to the slit, making the
image area entering the monochromator too small. A dove prism was used to rotate the image on
the entrance slit. A dove prism rotates an image twice as much as the angle that it is tilted
(Fig. 3-7).84,88 By positioning the dove prism between the two lenses at a 45 degree tilt, the
image was rotated by 900. The improvement in signal collection by using a dove prism is
illustrated in Figure 3-8.
The Raman signal is inherently weak due to a small Raman cross section in the order of
10-29 CM12 ST-1. It is difficult to detect a Raman signal amidst intense stray light, which causes an
overwhelming background. For this reason, many Raman instruments use double or triple
monochromators, which provide stray light rej section without sacrificing throughput.
Monochromators are dispersive instruments that isolate a small wavelength band from a
polychromatic source.84 Most monochromator configurations are based on the Czerny-Turner
design.84 Light passes through an entrance slit and reflects off a collimating mirror to a
diffraction grating. The grating separates the light according to wavelength. A focusing mirror
reflects the diffracted light onto an exit slit to the detector. It is important to take note of several
characteristics of the monochromator such as linear dispersion, resolution and spectral bandpass.
Linear dispersion shows the capability of the instrument to disperse light. It is how far
apart two wavelengths are in the focal plane. Linear dispersion is expressed as
DL = dx/dh = fDa (3 -2)
where f is the focal length of the focusing element in mm and Da is the angular dispersion in rad
nm- The unit for linear dispersion is mm nm-l Manufacturers of these instruments provide a
more common term, the reciprocal linear dispersion, RD which is given by
RD = (DL -1 = dh /dx = (FDa)^1 (3-3)
The reciprocal linear dispersion represents the wavelength range within a unit distance in the
focal plane and is conveniently expressed in units of nm mml
The size of the diffraction grating plays a key role in determining the solid angle, the f-
number, the throughput and the resolution. With a limiting aperture diameter of L in mm, a
projected area A and a focal length F in mm, the f-number (f/#) is expressed as
f/# = flL (3-4)
The solid angle 0Z in sr can now be defined as
0z = A/f2 = (2/4) / (f/#)2 (3-5)
The spectral bandpass, s,, is the half-width of the wavelength distribution that passes
across the exit slit. The spectral bandpass is expressed in units of nm. With the exception of
very small slit widths where diffraction effects and aberrations occur, the spectral bandpass for a
given slit width, W in mm is given by
s, = RDW (3 -6)
Generally, monochromators have equal entrance and exit slit widths. The slit widths determine
the spectral profie of the image at the exit slit. In the case of equal slit widths, when
monochromatic light is passed through the entrance slit, a monochromatic image of the entrance
slit is formed on the exit slit. As the grating is rotated, the entrance slit image is scanned across
the exit slit.
The monochromator' s resolution is closely related to its spectral dispersion. Dispersion
indicates the distance of 2 wavelengths in the focal plane while resolution specifies the
distinguishable separation of the two wavelengths. For larger slit widths where aberrations and
diffraction effects are negligible, spectral resolution is expressed as the slit width limited
resolution, ahs in units of nm
Ahs = 2s, = 2RDW (3 -7)
Resolving power is another way to express how well a monochromator distinguishes two
adjacent wavelengths. Experimentally, this can be determined using the expression
Rexp = have/Smin (3-8)
where smin is the spectral bandpass at the minimum slit width. The theoretical resolving power is
calculated using the expression
Rth = have ad (3 -9)
Stray light in a monochromator is considered to be any light that passes that is outside
his where h, is the wavelength setting and s, is the spectral bandpass. The intensity of the light
of other wavelengths is referred to as the stray light level. Stray light level is one of the most
important specifications of a monochromator for many applications.
A Spex 1680B double monochromator with a focal length of 0.220 m is used in this
research. It is equipped with a 1200 grooves/mm grating. The double monochromator's
dispersion is 1.8 nm mm l, the resolution is 0.2 nm at 500 nm and the numerical aperture is f/4.
It can be scanned over a spectral range of 185-900 nm. The most important characteristics of the
double monochromator are summarized in Table 3-3.
The next criteria to consider are the monochromator's accuracy and speed in acquiring a
spectrum. The grating motor drive dictates the accuracy and speed of measuring a spectrum.
The double monochromator used has an accuracy of 10.4 nm and a repeatability of 10.2 nm.
Experimentally, the monochromator scan speed in the continuous mode was determined at by
scanning from 500 to 600 nm at different wavelength increments (Table 3-4). The results of
these measurements also provided the approximate rotation time of the grating at specific
increments which was considered in the synchronization of the scanning with the photon
counting detection. This is covered in the next section of this Chapter. Monochromators involve
scanning the spectral features of the optical signal in both continuous and step modes. As a
result, the measurement process is slower than that of a spectrograph with a multi-array detector
such as charged coupled detectors (CCD). These types of spectrometers operate in a fixed
grating position and directly acquire spectrum according to their dispersion as opposed to
scanning the grating in the case of monochromators. The Spex 1680B double monochromator
has an internal drive circuitry that permits coupling with a scan controller. The scan can be
programmed to be in continuous or step mode. The continuous scan mode can easily be
achieved by using an external control unit that commonly comes with the monochromator.
However, these types of scan controllers cannot be used in step mode, which is necessary for
PMT-photon counting detection. An alternative and more convenient approach to
monochromator scan control can be achieved using a PC peripheral and software. Interfacing of
the monochromator to a computer was done using a SPEX232/488 spectrometer control interface
(JobinYvon; Edison, NJ), which has a dual compatibility. It can communicate with the
spectrometer via an RS232 serial port or the IEEE 488. A Windows-based spectrometer control
software (JobinYvon; Edison, NJ) was used to control the monochromator. The continuous scan
mode was used for direct current measurements while the step scan mode was used for photon
A Raman spectrometer must be able to measure weak Raman signal in the presence of a
stronger Rayleigh or diffuse reflections at the laser frequency. Raman spectrometers must have
an outstanding stray light rej section, which can be achieved by using double or triple
monochromators. The stray light cut-off of a typical double monochromator sometimes is not
enough to address this issue. Thus, optical filters are used to minimize detection of undesired
radiation entering the spectrometer, which can overwhelm the weak Raman signal. There are
two general types of filters that are used in Raman spectroscopy: laser transmitting and laser
blocking filter.89 Laser blocking filters are further grouped into two types: notch filter and long
wavelength pass filter or cut-off filter. These filters are characterized by their transmission
spectra. The transmission spectra illustrated in Figure 3-9 show the specific applications of the
three different types of filters.89 Laser transmitting filter positioned between the laser and the
sample limits undesired light from the laser such as broadband spontaneous emission or plasma
lines from being detected. Such filters clean up the laser lines allowing only a single laser line to
reach the sample. The thin-film laser blocking filter is positioned between the spectrometer and
the sample to suppress the Rayleigh scatter so that the weak Raman signals can be accurately
measured. In systems where both laser transmitting and notch filters are in use, both Stokes and
anti-Stokes Raman scattering can be measured simultaneously. The long wave pass filter on the
other hand, only measures the Stokes Raman scatter. The advantage of using a long wave pass
filter over a notch filter is illustrated by their transmission spectra. Long wave pass filters have
the ability to transmit light close to the laser line. A graph comparing the optical density [-log
(Transmittance)] of the two types of filters is shown in Fig. 3-10,89 which illustrates the increase
in the edge steepness of a long wave pass filter relative to a notch filter. The steepness of the
edge results in a narrow transition region separating the laser line and the transmitting region of
the filter spectrum, allowing the observation of Stokes Raman shifts very close to the laser line.
The Ar+ laser used was equipped with a broadband optic instead of a prism so the output
was a mixture of all the lines of an Argon ion laser. A laser transmitting filter (MaxLineTM
Semrock, NY) was used to clean up the laser light before reaching the sample. This filter only
allowed 514.5 nm laser line to pass while blocking the other laser lines (Fig. 3-11). A
transmission efficiency of more than 90% prevented significant loss in laser power. The
Rayleigh scatter has higher intensity than the Raman scatter and can overwhelm the weak Raman
signal. A laser blocking filter (Chroma Technology Corp; Rockingham, VT), positioned
between the monochromator and the sample blocks the Rayleigh scatter. This research only
involved Stokes Raman scatter measurement so a long wave pass filter was chosen over a notch
filter. A steep 514.5 nm long wave pass filter was used to block the Rayleigh scatter and
improve Stokes Raman measurement. The position of the filter was optimized and the Rayleigh
scatter in both configurations was compared (Fig. 3-12). Rayleigh rej section was observed to be
more efficient when the filter was positioned very close to the monochromator entrance slit.
Two different high pass filters were compared for their effectiveness in blocking the Rayleigh
scatter (Fig.3-13). The Razor Edge filter (Semrock; Rochester, NY) proved to be more effective
than the H-filter (Chroma Technology Corp; Rockingham, VT). Comparison of the properties of
these filters is presented in Table 3-5.89,90
Optical detectors are grouped into three categories: thermal, photon and multichannel
detectors.84 Photon detectors and multichannel detectors are commonly used in Raman
spectroscopy. Photon detectors are based on the rate of photon arrival and the spectral response
varies with wavelength while mutichannel detectors can provide simultaneous detection of
dispersed wavelength.84 Several types of photon and multichannel detectors are available.
Photon detectors are grouped into different types of devices as photoemissive devices
(photomultipliers and phototubes), pn-junction devices (photodiodes and phototransistors),
photoconductive cells, and photovoltaic cells.91
Photomultiplier (PMT) tubes fall under the photon detector category and are commonly
used in Raman measurements. Photomultiplier tubes are low noise light detectors with high
gain. Single photon detection can be achieved over a spectral range of 180-900 nm.
Photons striking the photocathode of a PMT cause ej section of an electron by the
photoelectric effect. The electron directed to a series of dynodes or high voltage steps gets
amplified. These electrons are collected at the anode with the signal output in the form of a
pulse. Photomultiplier detection has two main types: analog and photon counting. Analog mode
looks at the DC portion of the pulse. The DC current is the sum of all pulses regardless of their
source. For this reason, noise is added to the signal and difficult to eliminate. Analog
measurement is useful for measurements of high intensity signals because the noise contribution
is negligible. For low intensity signal measurements such as Raman spectroscopy, photon
counting is preferred. In photon counting, the pulses are directed to an amplifier/current-voltage
converter and then to a discriminator. The discriminator sets the cut-off voltage to filter low
voltage noise pulses, thus higher signal-to-noise ratios (S/N) are obtained. Most of the noise is
generated after the cathode and within the dynodes. These low voltage pulses are filtered by the
discriminator unlike in analog where noise is tightly bound to the signal. PMT's have high
voltage requirements and changes in the supplied voltage do not affect the pulse count while the
analog signal is affected. Another advantage is that photon counting is not affected by the RC
time constant that is inherent in the electronics of analog systems
A side-on R-928 HammamatsuThl photomultiplier tube cooled to about -200 C was used
as a detector. Important characteristics of this PMT are summarized in Table 3-6. Pulse
amplitude from the PMT was estimated based on gain and rise time provided by the
manufacturer using the formula92
Amplitude (mV) = Q x G x Z / tR (3-10)
where Q is the elementary charge in C/e, G is the gain, 0Z is the impedance of the preamplifier in
ohms and tR is the rise time in ns. The output from a PMT is a current pulse, which travels down
a 50 ohm cable that is terminated by a 50 ohm input impedance of a pre-amplifier. For an R-928
PMT, the gain is 1 x 107 and the rise time is 2.2 ns. Using the above formula, the calculated
amplitude of the pulse from the PMT is 36.4 mV.
A snubber was connected to the PMT to improve the shape of the pulse for photon
counting.92 This is a network consisting of a short piece of 50 ohm coaxial cable which is
terminated with a resistor of less than 50 ohms. Ringing, which can cause multiple counts from
a single photon, is very common when using a PMT. This can be minimized by connecting a
snubber to the PMT prior to photon counting. The snubber network is a 10 inch coax cable
connected to a small 50 ohm potentiometer with terminating impedance ranging from 0 to 50
ohms. The other end of the coaxial cable was connected to the PMT together with the output
signal cable that was connected to the preamplifier. The output current from the PMT is divided
into the signal cable and into the snubber. Adjusting the snubber to 50 ohms attenuates the
signal by a factor of 2. No signal is reflected back to the anode unless the pot is set to less than
50. Reflections are delayed by the cable roundtrip time and sent out the signal cable. The round
trip time in the snubber cable can be adjusted so that the reflections cancel the signal ringing.
With the correct snubber cable length, the round trip time of the cable can approximately match
the period of the signal ringing and thus, the signal ringing is cancelled. For a 10-inch snubber
cable, the round trip time is about 5 s. The snubber network connected to the PMT improved the
shape of the pulse. However, signal ringing was not completely cancelled. In addition to the 50
ohms impedance of the amplifier, the snubber had a terminating impedance of 50 ohms. Taking
into account these two 50 ohm loads, the pulse amplitude was estimated to be 18.2 mV and a
preamplifier gain of25 yielded about 455 mV.
Small amplitude noise pulses result from thermal emissions of the PMT dynodes and
amplifier. A discriminator is used to block these low amplitude noise. Setting the correct
discrimination level is important in photon counting. To achieve a good S/N, the discriminator
level should be set above the noise amplitude and below the signal amplitude. Another reason
for setting the correct discriminator level is to reduce drift. Slight change in the PMT gain may
lead to drastic change in the photon counts if the discriminator level is within the center of a
signal pulse height distribution. The discrimination level can be determined by plotting a pulse
height distribution as depicted in Figure 3-14.92 Generally, the optimum discriminator level for a
photon counter is within the valley of the pulse height distribution. There is no concrete rule in
setting the discriminator value as it depends on the nature of the measurement. With high dark
counts, the threshold should be set at a lower value. For a cooled PMT where dark counts are
not high, a higher threshold should work. A pulse height distribution presented in Figure 3-15
was plotted for the PMT. A narrow discriminator window is set by adjusting discriminator 1 to 0
V and discriminator 2 to -5 mV. The two discriminators were connected to the same input. The
narrow window between the two discriminators was scanned across all pulse height and the
photon counts were recorded. The discriminator level was set to -70 mV which is slightly higher
than the lowest value. This setting was chosen in order to eliminate the signal ringing, which the
snubber failed to cancel.
The SR-400 gated photon counter (Stanford Research; Sunnyvale, CA) was operated
using continuous gating and so the photon counter was enabled throughout the entire count
period. Computer control of the photon counter was done through the GPIB interface. The
photon counter was controlled by a LabVIEW program. All the parameters such as
discriminator level, integration time and count period were set using the LabVIEW software.
For all photon counting measurements, the integration time was set between 1 s to 30 s.
The spectrum of benzene recorded in analog was compared with that obtained with photon
counting. Figure 3-16 shows that photon counting gave a better S/N than the analog mode.
The LabVIEW program was also used to acquire spectra. Two different computer
interfacing methods were employed to drive the grating at a stepping mode and control the
photon counter. Synchronization of the monochromator control with the photon counter was
necessary to achieve accurate collection of Raman spectrum. In contrast to scanning
continuously, the step mode permitted longer integration time while scanning when recording a
Raman spectrum. A timing diagram (Fig.3-17) was followed to achieve smooth and
synchronized measurement and data collection. For the monochromator, the total time for each
measurement is not only the time set for the grating to stop at a certain wavelength but also the
time required for the grating to rotate from one wavelength to another during scanning. The
rotation time was calibrated at different wavelength increment (Table 3-4). The photon counter
has an integration time, which is the length of time the counter is set to count photons, and a
dwell time which is the time between count periods. Total measurement time set for the photon
counter should be equal to the total measurement time that the monochromator stops in between
The data acquired by the LabVIEW program were the counts corresponding to each
count period. Each data count corresponds to a specific wavelength. Data processing was done
using Origin 7.5 (Northampton, MA). The height of the peaks was determined by taking the two
lowest points on each side of the peak of interest and fitting a linear function.
All spectra were recorded using wavelength increments of 0. 1 nm to 0.2 nm. In order to
confirm that no peaks were skipped using these increments during photon counting, a spectrum
of a mercury lamp was recorded using both analog and photon counting mode. Figure 3-18
indicates that none of the narrow mercury lines were skipped during the scanning in the photon
Evaluation of Performance of the Raman System
The lab-built Raman system was designed specifically for measurements of liquids in
cuvettes. Several solvent spectra were obtained as part of the system characterization. Using
both analog and photon counting modes, spectra of pure solvents including benzene, methanol,
ethanol, tetrahydrofuran (THF) and acetone were recorded. The spectra of the pure solvents in
analog mode are presented in Figure 3-19. For all solvents, the spectra showed Raman peaks
matching those in literature. For the purpose of S/N consideration, the signal was defined as the
average intensity value and noise as standard deviation of the peak intensity. A comparison of
the benzene spectra indicated an improvement in the S/N was observed from analog mode to
photon counting mode (Fig. 3-16).
The ability of the system to observe Raman scattering from analytes in dilute solution
was demonstrated by the spectra obtained from beta-carotene in different solvents (Fig. 3-20).
Raman spectra of dilute samples of beta-carotene in THF were also measured.
The first step toward the SERS of DPA on silver colloids was to obtain a normal Raman
spectrum of saturated DPA as part of the system characterization (Fig. 3-21). A huge
background was observed from the saturated DPA. To determine the source of this background,
the spectra of an empty cuvette, deionized water and 1 M KOH were recorded and compared
(Fig. 3-22). The cuvette gave no signal while the deionized water and 1 M KOH showed the
same shape of background. Deionized water was present in both samples. A spectrum of
deionized water in literature confirmed the same Raman spectrum.93
After the design, optimization and characterization of the Raman system, a broader and
deeper understanding of the system has been attained. It is clear that the lab designed Raman
system is capable of measuring Raman scatter from a variety of liquid samples. The next step is
to assess its usefulness in SERS detection of DPA in bacterial spores.
Figure 3-1. The Raman experimental set-up.
Table 3-1. Common lasers used in Raman
Spectroscopy (Adapted from Reference #3).
Wavelength (nm) Typical average
Pulsed or quasi-
CW and pulsed
457, 488, 514.5
406, 647, 752
Table 3-2. Nominal output power of the visible lines of the Argon ion laser (adapted from
Output Power (W)
Visible Single Line
488 496.5 514.5
520 530 540 550
470 480 490 500 510
Figure 3-2. The Argon ion visible lines measured using an OceanOpticsTM CCD spectrometer.
Argn in Lse
Figure 3-3. The experimental set-up used for measuring Argon ion laser profile and laser short
and long term fluctuations.
0 200 400 600 800 1000
T im e (s)
Figure 3-4. The short-term fluctuation of the 514.5 nm line of the Ar' laser (%/ RSD=7%).
0 1000 2000 3000 4000 5000 6000
T im e (s)
Figure 3-5. The long-term fluctuation of the 514.5 nm line of the Arg' laser (% RSD = 8%).
Figure 3-6. A schematic diagram of the 900 sampling geometry adapted for Raman system.
~ 4 LL~q~~ ~ 4~1~
Figure 3-7. A diagram illustrating (a) the position of the dove prism in the Raman system and
(b) the image rotation by a dove prism (adapted from reference # 88).
-no dove prism
41 0 600 800 1000 1200 1400 1600
Ram an Shift (cm -1)
Figure 3-8. A comparison of normal Raman spectra of saturated DPA illustrating the
improvement in signal intensity due to the image rotation using a dove prism.
Table 3-3. A summary of the important characteristics of the Spex 1680B double
Reciprocal Linear Dispersion
0.2 nm at 500 nm
n speed (nm/s)
Table 3-4. The monochromator scan speed at different increments.
Increment (nm) rotation time (s) sca
0.1 0.240 0.4
0.2 0.243 0.8
0.5 0.256 1.9
1.0 0.355 2.8
2.0 0.469 4.2
5.0 0.829 6.0
10.0 1.499 6.6
450j 475 50P0 25 550 575 000
525; 575- 625 675 725
430 480 530 500
Figure 3-9. Transmission spectra of different types of SemrockTM Optical filters used in Stokes
Raman Spectroscopy: (a) laser transmitting filter, (b) long wave pass filter, and (c)
notch filter (adapted from reference 89 ).
7T NolBch Design
810 615 L20 02?5 6930 635P 180 645 660s 655 060a
Figure 3-10. A comparison of the optical density plot of a long wave pass filter vs. a notch filter
(adapted from reference #89).
450 460 470 480 490 500 510 520 530 540 550
Figure 3-11. A spectrum of Ar' laser showing the different output wavelengths (blue). Inserting
a 514.5 nm MaxlineTM Semrock laser line filter between the laser and the sample all
wavelengths except 514.5 nm.
512 513 514 515 516
517 518 519 520
H filter_ fle
Figure 3-12. The decrease in the intensity of Rayleigh scatter peak (5 14.5 nm) using a high pass
filter that blocks light from 500-515 with a transmission efficiency of >90% placed at
I ,,--Semrock nRzor~dge 1e
513 513 5 514 514 5 515 5155 516
Figure 3-13. A comparison of the Rayleigh peak attenuation by two different long wave pass
Table 3-5. A summary of the characteristics of the Hammamatsu R-928 PMT.
Photocathode material Multi alkali
Window Material UV glass
Peak Wavelength 400 nm
Spectral Range 185-900 nm
Dynode Stages 9
Gain 1 x107
Anode to cathode voltage (max) 1250 V
Average anode current (max) 0.1 mA
Anode dark current (after 30 min) typical 3 nA
Anode dark current (after 30 min) max 50 nA
Rise time 2.2 ns
Transit time 22 ns
Single photon peak
o Disc 1
Pulse Height (mV)
Figure 3-14. A typical pulsed height distribution. Disc 1 and Disc 2 are the discriminators of the
photon counter. A window discriminator (Disc 2-Disc 1) is scanned with a Atd
interval to generate a pulse height distribution in order to determine the
discrimination level for the photon counter.
1s Iegraton ime00
2s ltegrabon bme ,0
150 140 130 120 110 100 9o so 70 605o4
30 -180 -160 -140 -20 -100 -80 -60 -4
Pulse Height (mV)
Figure 3-15. The pulse height distribution for the photon counting system for determining the
discrimination level. Inset is the blown up portion of the plot to clearly show the data
Raman Shift (cm-1)
500 1000 1500 2000 2500 3000 3500
Raman shift (cm-1)
Figure 3-16. Benzene spectra obtained using analog and photon counting mode illustrating the
difference in signal to noise.
Figure 3-17. A timing diagram used for the photon counter and monochromator interfacing and
synchronization. The integration time and dwell time of the photon counter were set
using the LabVIEW program. The monochromator stop time and rotation time were
set using the monochromotor controller.
-0 2nm Increm ent
|-01 nm Increm entI
540 545 550 555 560 565 570 575 580 585
Figure 3-18. Mercury lamp spectra obtained with (a) analog and (b) photon counting mode
showing the narrow mercury lines at 546.08, 576.96 and 579.07 nm. This
demonstrates that the monochromator scanning does not skip any lines even when set
to 0.2 nm increments.
600 700 800 900 1000 1100 1200 1300 1400 1500 1600
Raman ShiR (cm-1)
700 800 900 1000 1100 1200 1300 1400 1500 1600
Raman ShiR (cm-1)
600 700 800 900 1000 1100 1200 1300 1400 1500 1600
Raman Shift (cm-1)
Figure 3-19. Raman spectra of several pure solvents measured using the analog mode.
1000- beta-carotene + ethanol
600 700 800 900 1000 1100 1200 1300 1400 1500 1600
Raman Shift (cm-1)
600 700 800 900 1000 1100 1200 1300 1400 1500 1600
Raman Shift (cm-1)
Figure 3-20. Raman spectra of 1.0 x 10-4 M beta-carotene in (a) ethanol and (b) tetrahydrofuran.
The spectra of the solvents are shown and can be superimposed on the beta carotene
spectrum. The arrows represent the peaks attribute to beta-carotene.
600 800 1000 1200 1400
R am anShift (cm -1)
Figure 3-21. A normal Raman spectrum of saturated DPA in 1 M KOH.
16000 -1 -e m pty q u artz c uve tte
9 -deionized water
] 14000 -( -1 M KO H
4( 0 600 800 1000 1200 1400 1600 1800
Raman Shift (cm l)
Figure 3-22. Raman spectra of an empty cuvette, deionized water, and 1 M KOH.
STABILITY OF SILVER COLLOID AS SUBSTRATE FOR SURFACE ENHANCED
RAMAN DETECTION OF DIPICOLINIC ACID
Metal colloids (hydrosols) have been commonly used as SERS substrates. Creighton et
al were the first to report SERS of pyridine on silver and gold colloids.47,72 Their measurements
have been significant in demonstrating that SERS is associated with surface plasmon excitation
on the metal surface.72 Subsequent experiments which followed the first observation of SERS of
pyridine on roughened electrodes involved the use of these types of substrates. However, they
are less favorable for fundamental studies of the SERS phenomenon. Researchers shifted their
attention to the use of metal colloids since it is much easier to characterize their optical
properties particularly their absorption spectra, and to account for these properties in terms of
roughness in a precise way. The main advantages of using metal colloids for SERS
measurements are their ease of formation and manipulation for analytical studies. Chemical
reduction of simple salts of silver or gold with either sodium borohydride or sodium citrate is
most commonly used as a method of preparing these metal hydrosols.
These colloidal systems include particles whose diameters range from 1 nm to 100 nm.94
These metallic particles are usually negatively charged due to the adsorbed anions. Neutral
species added to the hydrosol can be adsorbed on the particles replacing the negative charges.
This disrupts the aggregative stability of the system and larger aggregates can form.68 It has
been shown many times that SERS requires partial aggregation and adsorption of the
analyte.28,30,32,48,49,68,95,96 Although, analytical applications of SERS have been reported, the use
for quantitative analysis has been problematic because of the poor precision of the SERS signal
due to irreproducible colloid preparation. In order to understand the factors affecting the
irreproducibility of the SERS response, it is important to study SERS under static conditions.
This present study evaluates important factors such as silver colloid aging, analyte-induced
aggregation and the stability of the sodium borohydride. Dipicolinic acid (DPA) which is a
signature of Bacillus spores, is used in this study.
Experiments and Methods
Reagents and Procedures
Aqueous solutions of analytical reagent grade sodium borohydride (2 x10-3 M) and silver
nitrate (1 x 10-3 M) were used for generation of silver colloids. The chemical reaction is:28
Ag' + BH4- +3H20+ Ag +H3BO3 +7/2 H 2(g) (4-1)
Hydrogen gas is produced by the reduction of silver ions as well as the slow reduction of water
by the sodium borohydride at room temperature.
BH4- + 3H20 H2BO3- + 4H 2 (g) (4-2)
The silver hydrosols were obtained by mixing sodium borohydride and silver nitrate solutions at
6/1 (v/v) volume ratio. Aqueous solution of dipicolinic acid (1.5 x 10-3M) was mixed to the
resulting silver colloid at a 1:1 ratio. A list of reagents is provided in Table 4-1.
Silver Colloid Characterization
Characterization of the silver colloids by absorption spectroscopy was performed using
an Agilent UV-Vis double beam spectrophotometer. A 1 mL sample from each batch of silver
colloids was used for all absorption measurements. Deionized water was used as the blank.
Scanning electron microscopy (SEM) was used to study the surface morphology of the
silver colloids. 50 CLL aliquots were spotted on sample stubs and air-dried overnight in a
desiccator. The SEM samples were coated thinly with gold by sputtering at 50 mA for 50 s.
SERS measurements were conducted using the Raman system described in Chapter 3
(Fig.3-1). Briefly, it consisted of a Spectra Physics BeamLok 2060 Ar+ laser and a SPEX 1680B
double monochromator. A thermoelectrically cooled Hamamatsu R-928 photomultiplier tube
coupled with a Stanford research photon counting system was used as a detector. A dove prism
tilted at a 450 angle is used to rotate the image making it parallel to the entrance slit of the double
monochromator. A LabVIEW program was used for photon counter control and data
acquisition. All measurements were done using a standard 1 cm x 1 cm quartz cuvette. SERS
spectra were recorded in 90-degree collection geometry. All SERS signals were normalized to
the measured laser power after obtaining the peak.
Results and Discussion
The most commonly used method of generating silver hydrosols for SERS measurements
is the chemical reduction of silver nitrate by sodium borohydride solution. The quality of the
SERS signal depends on a variety of factors such as reagent concentration, mixing procedures as
well as the incorporation of the sample.28,48,4 Because of the dependency of the silver colloid
quality on several factors, it is not surprising that SERS precision is poor. This study evaluates
these important factors in order to improve the reproducibility of SERS measurement of
Farquharson and co-workers observed the following band in the SERS spectrum of DPA
in water: 657, 815, 1008, 1382, 1445, and 1567 cm 1.97 Normal Raman and SERS spectra of
dipicolinic acid were recorded to observe these bands (Fig. 4-1). The peaks in the normal Raman
and SERS spectra of DPA are summarized in Table 4-2. The most prominent peak observed at
1013 cm-l was barely visible in the normal Raman of 7.5 x 10-3 M DPA in water.
The hydrosols obtained in this study were characterized by their absorption spectra.
Provided that the silver particles are smaller than the wavelength of light and approximately
spherical in shape and monodisperse, silver colloids are yellow with a single extinction band at
385 nm. This band is due to the resonant excitation of plasma oscillations in the confined
electron gas of the particles.68 Figure 4-2 shows the effect of aging the silver hydrosols on the
absorption spectra. A sample hydrosol was obtained at different times after preparation for UV-
Vis absorption measurement. All the spectra showed a single maximum peak at about 400 nm
which is characteristic of the surface plasmon resonance of particles that are roughly spherical in
shape. An increase in the absorbance due to an increase in the concentration of the silver
particles over time was observed. This indicated that more silver particles form in a span of 1
hour after the reaction started. It is also evident that aging the silver colloid affects the SERS
signal of DPA. Figure 4-2 illustrates that as a consequence of the increase in the concentration
of silver particles through aging of the hydrosol, the SERS response of DPA increased for the
first 30 min and was constant for longer times.
Silver colloids usually have a negatively charged surface due to adsorbed anions from the
reagents used in the reaction and electrostatic repulsion keeps the colloid stable from aggregation
provided that there is sufficient charge.68 Introduction of molecules such as DPA can replace
these negative charges leading to aggregation of the silver particles.68 Before addition of DPA,
the silver colloids were yellow in color with a sharp extinction band centered at 400 nm.
Addition of DPA changed the color of the silver colloid solution from yellow to pink indicating
the formation of aggregates. Aggregate formation was evident from the change in the shape of
the silver hydrosol absorption spectrum with the appearance of another peak at longer
wavelengths (Fig.4-3). No drastic changes in the SERS signal of DPA were observed when the
DPA and the silver colloid mixture was allowed to stand for 1 h. Therefore, further silver
colloid aggregation induced by the DPA was minimal which could also be deduced from the
observed absorption curves of the mixture monitored within the 1 h period.
Sodium borohydride has been commonly used as a reducing agent for the generation of
silver hydrosols. Equation 4-2 shows that further oxidation of the borohydride ions occur even at
room temperature. To avoid such concentration changes over time, sodium borohydride
solutions must be freshly prepared. Figure 4-4 demonstrates the depletion of borohydride ions
available to reduce silver nitrate over a period of 1 h. Assuming that all silver nitrate reacted
with sodium borohydride, a decrease in the absorbance at the maximum peak ca. 400 nm
indicates a decrease in the resulting silver colloid concentration. The absorbance observed after
1 h of standing the borohydride solution decreased two-fold as shown in Figure 4-4a.
Deoxygenation of the deionized water was done prior to preparation of sodium borohydride
solution to avoid further oxidation to occur. Purging the solution with an inert gas such as
nitrogen is the most commonly used method of minimizing oxygen in the solution. However,
this method did not completely stop the further oxidation of borohydride based on the
absorbance at the peak maximum spectra of the silver hydrosols (Fig. 4-4b). Another way to
address this problem is to introduce another reducing agent such as sodium sulfite. Sodium
sulfite is used as an oxygen scavenging agent in water treatment. This method has been used to
prevent quenching of phosphorescence by oxygen.98 This method is based on the reaction
2SO3-2 + 02 + 2SO4-2 (4-3)
This deoxygenation method showed better results as no drastic decrease in the absorbance at the
maximum peak was observed (Fig. 4-4c). The resulting silver concentrations remained constant
with the addition of sodium sulfite. In Figure 4-5, an improvement in the SERS signal
reproducibility was obtained largely a result of limiting borohydride concentration changes with
the aid of the sodium sulfite addition.
It was also observed that the formation of the yellow silver colloid solution, upon
addition of sodium borohydride to silver nitrate occurred at a slower rate when sodium sulfite
was added. In the absence of sodium sulfite, the color change is instantaneous. A kinetic study
of the silver colloid formation was carried out over a period of 3 h. A comparison of 2 different
concentrations of sodium sulfite is presented in Figure 4-6. The absorbance reading reached a
plateau after 50 min and 75 min for the silver colloid treated with 0.005 M and 0.01 M Na2SO3,
respectively. The stability of SERS signal observed from DPA using silver colloids treated with
0.050 M Na2SO3 after 50 min (Fig. 4-5a) was consistent with the results obtained from the
absorption measurements of the silver. SEM images of the silver colloids collected at different
times are shown in Figure 4-7. No change in the particle size was observed from each silver
colloid. Comparison of the SERS signal obtained using silver colloids treated with 0.005 M and
0.010 M Na2SO3 yielded different signal enhancement factors (Table 4-3). Two orders of
magnitude enhancement was achieved using 0.005 M Na2SO3 and one order of magnitude with
0.01 M Na2SO3. Sulfate aggregation has been shown to promote SERS of DPA.99 Based on the
equation 4-3, sulfate induced aggregation may occur after the oxidation of sulfite. Addition of
0.005 M Na2SO3 WAS sufficient to form aggregation resulting in a 100-fold DPA SERS signal
enhancement while 0.010 M resulted in excess aggregation, hence yielding only an order of
magnitude enhancement. It was also observed that sulfite addition gave a higher DPA SERS
signal as compared to sulfate addition as summarized in Table 4-3.
This preliminary study of the SERS of DPA is crucial in improving the reproducibility of
SERS as an analytical technique. It is important to understand the SERS of DPA on silver
colloids in static conditions to address the problem of poor precision. It was shown that the age
of the silver colloid affects the observed SERS signal of DPA with the optimum signal achieved
using silver colloids aged for at least 30 min. The DPA-induced aggregation study shows that
the SERS signal measured over 1 h was stable allowing longer measurement time, as in the case
of scanning PMT-photon counting detection. Control over the depletion of borohydride ions
available for reduction of silver ions was achieved by using sodium sulfite for deoxygenation.
All these findings are useful in the development of a Flow Inj section Analysis (FIA)-SERS
system for the detection of DPA which is the subj ect of the Chapter 5.
Table 4-1. List of reagents used.
Reagent Systematic Formula Molecular Physical Company
name weight state
Dipicolinic Pyridine-2,6- (C5H3N)(COOH)2 167.12 White Sigma-
acid dicarboxylic powder Aldrich
Sodium Sodium NaBH4 37.83 White Fisher
borohydride tetrahydroborate powder Chemicals
Silver nitrate Silver nitrate AgNO3 169.8731 White Fisher
Sodium Sodium sulfite Na2SO3 126.04 White J.T Baker,
sulfite crystalline NJ
Sodium Sodium NaCl 58.44 White Fisher
chloride chloride cry stalline Chemi cal s,
Sodium Sodium sulfate Na2SO4. 10 H20 322.19 White Fisher
sulfate decahydrate crystalline Chemicals
Nitric Acid Nitric acid HNO3 63.01 Liquid Fisher
Potassium Potassium KOH 56.11 White Fisher
hydroxide hydroxide pellets Chemicals
4* 0 600 800 1000 1200 1400 1600 1800
Raman Shift (cm-1)
Figure 4-1. Comparison of the normal Raman spectra of (a) saturated DPA in KOH and (c) 7.5 x
10-3 M DPA in water; and (b) SERS spectrum of 7.5 x 10-3 M DPA in water. The
silver colloids used for the SERS measurement were prepared by chemical reduction
of 1 x 10-3 M AgNO3 by 2 x 10-3 M NaBH4. The volume ratio used was 6 to 1
(NaBH4 to AgNO3).
Table 4-2. A summary of observed normal Raman and SERS peaks of DPA.
Normal Raman SERS Normal Raman SERS Peak Assignments
647 657 674 674
Symmetric ring stretch
l i l ii l II I .20
35000 .- -- 0.18
o 25000- -I 0.14
20000- ----- m 01
m 0.10 0
O'' 10000 -0.06
0 0 ~~....0.00
2 0 10 20 30 40 50 60
Tim e (m in)
Figure 4-2. SERS signal of DPA on silver colloids aged over 1 h and the absorbance of the
silver colloid at 402 nm measured from 0 to 60 min. The silver colloids used were
prepared by chemical reduction of 1 x 10-3 M AgNO3 by 2 x 10-3 M NaBH4. The
volume ratio used was 6 to 1 (NaBH4 to AgNO3 -
0.5 -I | liver loniola + v ey
300 400 500 600 700 800
Wavenum ber (cm l)
Figure 4-3. Absorption spectra of silver hydrosol before (black) and after (red) addition of
dipicolinic acid (DPA). The silver colloids were prepared by chemical reduction of 1
x 10-3 M AgNO3 by 2 x 10-3 M NaBH4. The volume ratio used was 6 to 1 (NaBH4 to
o I I ~no treatment
Q ~ --ith n itrog en p urgin g
0.2-c with sodium sulfite a -
0.0 ggggg .
O 10 20 30 40 50 60
Figure 4-4. Absorbance of silver colloids at the peak maximum 401 nm as a function of sodium
borohydride aging over a period of 1 h: (a) no treatment, (b) with nitrogen purging
and (c) with sodium sulfite. The silver colloids were prepared by chemical reduction
of 1 x 10-3 M AgNO3 by 2 x 10-3 M NaBH4. The volume ratio used was 6 to 1
(NaBH4 to AgNO3).
SI I I I I
0 2D AD O ED 1D ~120
SERS signal of DPA on silver colloid with (a) and without sodium sulfite as oxygen
scavenger (b The silver colloids used in the SERS measurements were prepared by
chemical reduction of 1 x 10-3 M AgNO3 by 2 x 10-3 M NaBH4. The volume ratio
used was 6 to 1 (NaBH4 to AgNO3).
(11 0 7-
4I m 0.005 M Na2SO3
a 0.010 M Na2SO3
0 2000 4000 6000 8000 10000 12(000
Figure 4-6. Absorbance of the silver colloids at the peak maximum 402 nm. The silver colloids
were prepared by chemical reduction of 1 x 10-3 M A~gNO3 by 2 x 10-3 M NaBH4
stabilized with 0.005 M and 0.010 M Na2SO3. The volume ratio used was 6 to 1
(NaBH4 to AgNO3).
Figure 4-7. SEM images of silver colloids prepared using sodium borohydride treated with
sodium sulfite. Silver colloid morphology was observed (a) 15 min and (b) 30 min
after preparation. The silver colloids were prepared by chemical reduction of 1 x 10-3
M AgNO3 by 2 x 10-3 M NaBH4 Stabilized with 0.005 M Na2SO3. The volume ratio
used was 6 to 1 (NaBH4 to AgNO3).
Table 4-3. A summary of normalized signal intensities and calculated enhancement factors from
SERS of DPA on silver colloids.
Normalized Intensity (Counts/W)
DPA Silver Colloid
Concentration Treatment Enhancement
Normal Raman SERS Factor
7.50E-03 0.005 M Na2SO4 38253 71
7.50E-03 0.010 M Na2SO4 44039 82
7.50E-03 0.005 M Na2SO3 54062 101
7.50E-03 0.010 M Na2SO3 5857 11
EVALUATION OF EXPERIMENTAL CONDITIONS FOR THE SURFACE ENHANCED
RAMAN DETECTION OF DIPICOLINIC ACID ON SILVER COLLOIDS GENERATED BY
FLOW INJECTION ANALYSIS
Flow inj section analysis (FIA) is based on the inj section of a plug of sample into a moving
and continuous carrier liquid. The injected sample forms a zone and is transported to a detector
which continuously records a signal as the dispersed sample passes through a flow cell. The
simplest FIA system is composed of a pump to push the carrier liquid through the small diameter
tubing, an injection port for reproducible injection of the sample to the carrier stream, a reaction
coil for sample dispersion and reaction with the carrier liquid, and a detection flow cell. An FIA
output shows that peak height is related to the analyte concentration. A well designed FIA
system has an extremely rapid response in the range of 5-20 s with injected volumes between 1
CLL to 200 CLL.101 This makes FIA a high throughput technique with minimum reagent and sample
requirement. FIA systems are designed for the purpose of analyzing the maximum number of
samples at short time and minimum reagent and sample volumes.
FIA is governed by three basic principles as described by Ruzicka and Hansen.102 These
are reproducible timing, sample injection and controlled dispersion. The purpose of transporting
sample solution using an FIA is to treat the sample material in such a way that it will yield not
only a sensitive output but also reproducible. In an FIA system, physical mixing of the reagent
and sample is carried out reproducibly. The use of an injection valve prevents variation in the
sample volume. In such systems, band broadening or sample zone dispersion is controlled to suit
the detection and chemistry associated with it.
The process of dispersion, as the sample plug is transported through FIA system, can be
quantitated using the dispersion coefficient. Dispersion is defined as the amount by which the