Surface enhanced raman spectroscopy on metallic colloids for the purpose of trace analysis

MISSING IMAGE

Material Information

Title:
Surface enhanced raman spectroscopy on metallic colloids for the purpose of trace analysis
Physical Description:
ix, 138 leaves : ill. ; 28 cm.
Language:
English
Creator:
Angebranndt, Martin John, 1964-
Publication Date:

Subjects

Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1991.
Bibliography:
Includes bibliographical references (leaves 132-137).
Statement of Responsibility:
by Martin John Angebranndt.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001693374
notis - AJA5453
oclc - 25222112
System ID:
AA00003711:00001

Full Text
















SURFACE ENHANCED RAMAN SPECTROSCOPY ON METALLIC COLLOIDS
FOR THE PURPOSE OF TRACE ANALYSIS














BY

MARTIN JOHN ANGEBRANNDT


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


1991














ACKNOWLEDGEMENTS

I would first like to thank Dr. James D. Winefordner for

his guidance and support. I am especially grateful for the

considerable freedom he allowed me, while keeping me from

straying too far off course. I am also grateful to my

colleagues in Jim's laboratory and throughout the department

for all of the good and bad times we shared. I hope

Washington D.C. is not too far out of the way. Another thank

you must go to Tony Antonazzi, who operated the photon

correlation spectrometer for the experiments on the copper

colloids.

Considerable recognition must also go to Frangois Lang,

Bob Wallace, Ward Campbell, Dr. Robert Ho, and John

Glendenning, my Aikido Senseis, as well as the other Aikidoka.

They managed to keep my physical and spiritual development

near the mental and technical levels of my academic pursuits.

Finally, I am grateful to my family. My parents and

sister, who helped and supported me when I needed it most and

my new wife, who helped me keep a smile during one of the most

trying periods in my life.
















TABLE OF CONTENTS


Pace


ACKNOWLEDGEMENTS .

LIST OF TABLES . .

LIST OF FIGURES . .


ABSTRACT

Chapter


INTRODUCTION .


OVERVIEW OF PREVIOUS WORK .

SERS on Electrochemical Surfaces
Analytical Application .
Physical Studies .
SERS on Colloidal Metals .
Analytical Applications .
Physical Studies .
SERS on Chromatographic Surfaces

THEORETICAL FOUNDATIONS .


. 20


The Raman Scattering Process .
The SERS Enhancement Process .
Electromagnetic Enhancement Theory .
Chemical Enhancement Theories .

COLLOIDS AND AGGREGATION .

SILVER COLLOIDS OF SULFA DRUG ANALYSIS

Introduction . .
Experimental . .
Apparatus . .
Chemicals and Procedure .
Results and Discussion . .


iii


rrii


. vi


viii


42

50









6 A STUDY OF COPPER COLLOIDS . 68

Introduction . 68
Experimental . 70
Apparatus . .. 70
Chemicals and Procedure .. 73
Results and Discussion .. 73

7 COPPER COLLOIDS IN A FLOWING STREAM 95

Introduction . ... 95
Experimental . ... 97
Apparatus . .. 97
Results and Discussion . .. 100

8 CONCLUSIONS AND SPECULATIONS ON THE FUTURE 111

Conclusions . . 111
Speculations . 113
Experiments on Filters ... 113
Novel Colloidal Surfaces .. 115
Methodology .. . 115

APPENDIX ELECTROMAGNETIC DERIVATIONS .. 117

Molecule Near a Hemispheroid Protruding
From a Surface .. . 117
Full Prolate Spheroid .... 126
The Enhancement Factor ... 128

REFERENCES . . 132

BIOGRAPHICAL SKETCH . . 138
















LIST OF TABLES


Page

5.1 Intensities and Positions of Raman Bands
for the Sulfa Drugs as KBr Pellets 57

5.2 Intensities and Positions of SERS Bands
of Sulfa Drugs on Silver Colloids ...... 60

5.3 Analytical Figures of Merit for Sulfa
Drugs on Silver Colloidal System. ... 66

6.1 Analytical Figures of Merit for Analytes on Static
Copper Colloid . .. 93

6.2 Comparison of Static p-aminobenzoic acid Results
with Literature . .. 94

7.1 Effect of Acidity on SERS Signals and Colloidal
Aggregation . .. 102

7.2 Comparison of p-aminobenzoic acid FIA Data with
Literature . .. 109

A.1 List of Symbols used in the Appendix 131
















LIST OF FIGURES


Page

3.1 A Schematic Diagram showing the Stokes and
Anti-Stokes Shifted Raman Scattering Relative to
the Rayleigh Scatter and their Origin 22

3.2 The Real and Imaginary Components of the Dielectric
Functions of Copper and Silver. .. 29

3.3 A Schematic Diagram Illustrating the Difference
between the Normal Raman and the SERS Processes 31

3.4 A Plot Showing the Relative SERS Enhancement for
Silver Spheroids as a Function of Polar Angle 34

3.5 A Plot Showing the Effect of Separation Distance on
the SERS Enhancement for Silver Spheroids 36

4.1 Calculated Extinction Profiles for 20 nm spheres of
Copper and Silver in Water . .. 44

5.1 Normal Raman Spectra of the Sulfa Drugs as
KBr Pellets . . .55

5.2 SERS Spectra of the Sulfa Drugs on
Silver Colloids . .. 58

5.3 The Structures of the Sulfa Drugs .. 62

5.4 The SERS Spectrum of Sulfamethazine at 1, 10,
and 100 ppb Levels . 64

6.1 A Schematic Diagram of the Experimental Layout of
the Optical Components Used in the Copper Colloid
Studies . . ... .71

6.2 Normal Raman Scattering for Pyridine, SERS Spectrum
of 10 ppm Pyridine on a Copper Colloid, and the
Background of the Copper Colloid ... .75

6.3 Relative SERS Intensity of the Most Intense Band of
the Analyte as a Function of pH .. 77








6.4 Absorption Scans of the Copper Colloid as it
Aggregates over Time with PABA . 80

6.5 A Photon Correlation Spectrum of the
Copper Colloid . ....... .82

6.6 Intensities of the Two Most Intense Bands of PABA as
a Function of Time .............. .84

6.7 SERS Spectrum of 50 ppm PABA on a Copper Colloid
and the Background of the Copper Colloid ... 86

6.8 SERS Spectrum of 50 ppm PNBA on a Copper Colloid
and the Background of the Copper Colloid .. 88

6.9 SERS Signal as a Function of Analyte Concentration
for Pyridine, PABA, and PNBA .. 90

7.1 Schematic Diagram of the FIA Setup used to Produce
the Copper Colloids. . 98

7.2 The SERS Spectrum of PABA of a FIA Produced Copper
Colloid . . 103

7.3 The SERS-FIA Signals at 1397 cm'1 from a Series of
200 microgram injections of PABA into the System 105

7.4 SERS Signal as a Function of Amount of
PABA Injected . .. .. 107

A.1 a). A Cross-section of a Prolate Spheroid
Illustrating the Elliptical Coordinate System.
b). An Illustration of the Model used for the
Derivation in the Appendix . ... 119


vii


















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

SURFACE ENHANCED RAMAN SPECTROSCOPY ON METALLIC COLLOIDS
FOR THE PURPOSE OF TRACE ANALYSIS


BY

MARTIN JOHN ANGEBRANNDT

August, 1991



Chairman: Dr. James D. Winefordner
Major Department: Chemistry



The use of Surface Enhanced Raman Spectroscopy (SERS) on

silver and copper colloidal surfaces is discussed.

Qualitative and quantitative results in the determination of

three sulfapyrimidine drugs on silver surfaces are presented.

It is demonstrated that it is possible to differentiate

between these drugs at trace levels. Limits of detection in

the low ppb range were obtained with a linear dynamic range

extending for approximately two orders of magnitude above this

for the 1114 cm1 band.

The analytical figures of merit of copper colloidal

surfaces are evaluated for the first time. These copper


viii









colloids are produced without the addition of protecting

agents. The only previous preparation used citrate as an

antiaggregant. These colloids are less stable than silver

colloids, but will have fewer interference from fluorescent

solution species. Analytical results for p-aminobenzoic acid,

p-nitrobenzoic acid, and pyridine adsorbed onto a batch

produced colloid are presented. The detection limits are in

the low ppm range. This is comparable with previous results

for silver colloidal surfaces.

A Flow Injection Analysis (FIA) system was constructed to

evaluate the possibility of adapting copper colloid based-SERS

to on-line analysis. It was possible to continuously produce

the colloid and analyze a sample injected into the flowing

stream. Para-aminobenzoic acid was used as the test analyte.

The detection limit was found to be 150 gg. The major

obstacle in this analysis is the hydrogen gas produced as a

byproduct of the reduction. The gas forms bubbles in the line

and must be removed prior to the spectroscopic measurement.

The method used to remove the hydrogen gas is just adequate,

and an improvement in this aspect of the analysis would

increase the reproducibility significantly.














CHAPTER 1

INTRODUCTION

Raman Spectroscopy provides information which cannot be

obtained easily by other means [1]. It allows the researcher

to probe the symmetric vibrations of a molecule. This

information is complementary to that gleaned using infrared

spectroscopy. The primary reason that Raman spectroscopy can

be such a powerful tool for analytical analysis is that all

molecules possess some vibrations which are Raman active.

Therefore, this method can be applied when the various

luminescence spectroscopies cannot. The Raman spectrum also

contains more information than a luminescence spectrum, enough

to identify the molecule.

The major disadvantage of Raman spectroscopy is the low

quantum efficiency of the process [2]. Since only a very few,

approximately 1 in 10+6, of the photons are Raman scattered,

an intense excitation source is required to observe a signal

within a reasonable period of time. The signal observed is

located close to the excitation wavelength requiring the

excitation source to be also spectrally narrow, as well as

intense. A laser fits both of these qualifications and is

responsible for the increased interest in Raman Spectroscopy

since the invention of the laser in 1964. A double or triple









2

monochromator or a notch filter is needed to reject the

intense laser radiation, so that small Raman signals near the

laser line can be observed.

Because a Raman signal can be generated by any laser

wavelength, the researcher can choose the spectral region he

wishes to work in [3]. The ability to monitor the signal in

the visible or ultraviolet regions allows vibrational

spectroscopy to be carried out in aqueous solution, which is

difficult with infrared absorption. It also allows one to

choose a portion of the spectrum which is free from other

spectral interference, such as fluorescence from solution

species [4]. Equipment already existing in the laboratory for

other types of analysis, such as luminescence, can be adapted

to monitor Raman signals. It is also relatively easy to probe

vibrations below 600 cm1, which is difficult with IR

absorption spectroscopy because of the poor sources, optics,

and detectors available for the far infrared region.

One of the methods for increasing the quantum efficiency

of the Raman process is to adsorb the molecule of interest

onto a surface [5]. Surface Enhanced Raman Spectroscopy

(SERS) involves both electromagnetic enhancements in the local

field and the chemical enhancements in the molecular

polarizability to increase the intensity of the Raman scatter.

Several surfaces, most notably gold, silver and copper,

provide enhancements of up to six orders of magnitude. This

enhancement over the Raman scatter from the free molecule










opens the door for trace analysis using Raman spectroscopy.

The nature of the enhancement mechanism is dependent on the

orientation of the molecule relative to the surface to be

probed. The enhancement and the orientational information

combine to yield a probe of startling spectral and spatial

resolution. A limitation of this method is the necessity of

close interaction between the molecule and the surface. This

limits the analytical application of this technique to

molecules with functional groups containing nitrogen, sulfur,

oxygen and phosphorus.














CHAPTER 2

OVERVIEW OF PREVIOUS WORK

SERS on Electrochemical Surfaces

SERS was first observed on a silver electrode by

Fleishmann et al. [6] who were interested in using Raman

spectroscopy to study species adsorbed at the electrode

surface. Intense Raman scattering was observed from pyridine

adsorbed at the roughened silver electrode surface. They

attributed the high intensity of the Raman scatter from

pyridine to an increase in the number of molecules able to be

adsorbed onto the roughened surface.

Later Fleishmann et al. [7] investigated pyridine

adsorbed onto a copper electrode, but failed to see an

increase in the Raman scatter because they used 514.5 nm for

the excitation when red wavelengths were required. They did,

however, observe an increase in the adsorption of pyridine to

the electrode surface with increased chloride concentration,

the subject of many later studies.

SERS was later observed on copper and gold electrode

surfaces by Wenning et al. [8,9]. They reported that SERS was

only observed on gold and copper surfaces when irradiated by

"red" wavelengths. The observed enhancement in the intensity

of Raman scatter from pyridine was five orders of magnitude









5

for a copper electrode surface and four orders for a gold

electrode surface. These studies included measuring the

angular dependence of the SERS signals on electrode surfaces.

The optimum collection angle for SERS measurements was found

to be 550 from the normal of the surface.

Jeanmarie and Van Duyne [10] and Albrecht and Creighton

[11] independently investigated the silver electrode-pyridine

system. Interpreting the results within the framework of

experiments by Barradas and Conway [12] on the adsorption of

pyridine onto metal surfaces, they concluded that the increase

in the intensity of the Raman scatter was not due to increased

adsorption. Both groups based this on the fact that a single

roughening cycle of the electrode yielded a greater

enhancement than multiple roughening cycles. The increase in

the Raman scattering was correctly attributed to an increase

in the Raman cross-section due to the adjacent rough metal

surface, rather than an increase in number of molecules

absorbed onto the surface because of the increased surface

area of the electrode.

Analytical Applications

Carrabba et al. [13] evaluated the potential of SERS on

silver electrodes for determinations of organic water

contaminants. All of the experiments in this study were

performed in low ionic strength solutions to simulate in situ

groundwater conditions. No interference was observed from

humic acid materials, which would be present under field









6
conditions. The silver electrode potential controlled the

adsorption of pyridine and quinoline. Variation of the

potential allowed discrimination between these two compounds.

The detection limit for pyridine was estimated to be 8.5 pg.

Force [14] reported the first analytical application of

SERS at an electrode surface to a flowing system. The silver

electrode underwent a short Oxidation-Reduction Cycle (ORC)

prior to sample introduction. The electrode potential was

then adjusted so that the pyridine would adsorb onto the

surface. After the spectrum was obtained, the potential was

adjusted to desorb the pyridine and another short ORC

performed. This approach obviously can only be used for

analytes which adsorb reversibly onto the surface. A detection

limit of 250 nmol was obtained with linearity extending for

three orders of magnitude above the detection limit. Pothier

and Force [15] have also used this system for the analysis of

adenine, thymine and cytosine. They obtained detection limits

of 175, 233, and 211 pg, respectively, for these compounds,

using a multichannel analyzer. The linear dynamic range

extended four orders of magnitude above these values.

Cotton et al. [16] combined SERS with a resonance

enhancement performing SERRS on a silver electrode surface.

This technique was used to detect four nitrophenol compounds

previously separated by HPLC. The eluent was collected by a

fraction collector, and the spectra measured off-line using a









7

silver electrode surface. Detection limits of approximately

20 ppb were reported.

Physical Studies

The SERS spectrum of pyridine on copper, silver, and gold

electrode surfaces has been the subject of many

investigations. The excitation profile of the pyridine

breathing mode for each of these surfaces was reported and

discussed in terms of a chemical enhancement mechanism [17].

Changes in the relative intensities, frequency shifts and

increases in the half-width were indicative of metal-adsorbate

complexation. Increases in the intensity of the scattering

were observed when chloride ions were present, indicating that

chloride ions mediated the surface complex formation. They

also suggested that most of the surface features are smaller

than 5 nm, since a stronger oxidation-reduction cycle (ORC)

would be required for the formation of larger surface

features.

The pH dependence of the SERS signal of pyridine adsorbed

onto a copper electrode was measured by Cooney and Mernagh

[18]. They reported that an ORC was not necessary for the

observation of SERS from a copper electrode, as it was with

silver and gold surfaces. They attributed this to the strong

coordination of pyridine with the surface. The optimum pH for

the observation of SERS from the pyridine copper electrode

system was found to be = 7 Evidence of laser damage to the

electrode surface by the krypton ion laser also was discussed.








8

Crookell et al. [19] used an FT-IR spectrometer to

measure SERS in the near infrared region. The 1064 nm line of

a Nd-YAG laser was used to irradiate copper, silver and gold

electrode surfaces, with two narrow bandpass filters being

used to excluded the laser scatter. SERS signals were

observed from pyridine adsorbed on these surfaces. The

potential dependence of the SERS signal for the pyridine-

silver electrode system from -0.1 to -0.9 V vs. SCE was

reported. It was noted that this might be an ideal region,

since fluorescence was not often a problem, and it was still

possible to obtain high quality Raman spectra from aqueous

systems.

The SERS spectra of the adenosine and two of its

derivatives on a silver electrode surface were reported by

Koglin et al. [20]. They found indications that the w bonds

of the ring systems was able to interact with the surface,

giving rise to SERS. It was noted that the presence of (PO3)2

groups in a molecule significantly increased the Raman

intensity, due to their strong association with the surface.

The SERS spectra of adenine, guanine, thymine, and

cytosine on silver electrodes were reported and compared by

the same group [21]. Differentiation between these DNA bases

is possible from the frequencies of the breathing modes alone.

A partial assignment of the bands was also presented. A more

complete assignation of the bands arising from these DNA bases









9
and a discussion of the orientation of these molecules on the

surface was given by Otto et al. [22].

SERS on Colloidal Metals

SERS on colloidal metal surfaces was first observed in

1979. Creighton et al. [23] reported intense Raman scattering

from pyridine on colloidal silver and gold surfaces. The

excitation profiles for pyridine on both of these surfaces was

reported. A correlation between the extinction profiles of

the hydrosols, and the excitation profiles of pyridine on

these surface was shown. This was the first paper to assert

that the SERS process involves the excitation of surface

plasmons. Both of these colloids were prepared using a

borohydride reduction.

Creighton et al. [24] were also the first to observe SERS

with copper colloids. A borohydride reduction was again

used, but citrate was added to stabilize the colloid. The

SERS spectrum of pyridine, diphenyl sulfide, thiophenol and

4,4'-bipyridyl on this surface were reported. In a mixture of

pyridine and another of the analytes, competitive adsorption

was observed at the surface. A Raman excitation profile

similar to that of gold was obtained for pyridine. The

maximum enhancement observed was calculated to be 1.5 X 105,

50% of that of that for a similar gold hydrosol.











Analytical Applications

Torres [25] was first to use silver colloids for

analytical trace analysis. SERS spectra for p-aminobenzoic

acid, phenytoin, 2-aminofluorene, as well as uracil and some

of its derivatives, were reported. The effects of different

preparations on the analytical figures of merit of 2-

aminofluorene, and p-aminobenzoic acid were also studied. All

of the colloids were produced in batches (static conditions).

A later study by Morris et al. [26] used sedimentation to

fractionate silver colloids prior to the analysis. Crystal

violet was used as the analyte to evaluate the SERS activity

of these fractions. The linear dynamic range for the fraction

yielding the greatest enhancement was between 1 pg and 1 ng.

This fraction was only partially aggregated. The use of this

fractionation technique increases the amount of SERS active

surface available for the analyte to adsorb onto relative to

the total surface area of the colloid.

The use of photoreduced silver as a SERS substrate was

evaluated by Ahern et al. [27]. The silver metal was produced

via a photoreduction of silver nitrate. The spectra were good

relative to chemically produced colloids, exhibiting better

signal to noise characteristics than the silver hydrosols

produced in that study. It was not possible to observe SERS

from species which are highly soluble in water. There was

some problem with preparing a narrow distribution of particle

sizes.









11

Silver colloids were used in the trace analysis of sulfa

drugs by Sutherland et al. [28]. Limits of detection were in

the low ppb range, with the added ability to qualitatively

differentiate between sulfadiazine, sulfamerazine, and

sulfamethazine.

Laserna et al. [29] first applied colloidal SERS to Flow

Injection Analysis (FIA). For para-aminobenzoic acid (PABA),

the limit of detection was in the ppb, while a 5% RSD was

obtained. Their research demonstrated that the variance in

colloidal SERS intensities was due mainly to variation in

colloid preparation. FIA allowed the aggregation of the

hydrosol, the mixing of the analyte and the colloid, and the

time of observation to be precisely controlled. The effect of

pH on the SERS signals of PABA was also investigated. The

SERS spectrum of 9-aminoacridine [30] was later evaluated to

determine the applicability of this technique to other

systems. The effects of the addition of various ions was also

investigated.

Pararosaniline was used as the analyte by Freeman et al.

[31] to test an adaptation of colloidal SERS to HPLC. In this

paper, a previously prepared silver hydrosol was mixed with

the eluent from an HPLC. A linear range between 0.1 and 50

ppm was reported for this dye.

FIA was used by Taylor et al. [32] to evaluate combined

resonance and surface enhancement in amplifying the Raman

signal (SERRS) for application to HPLC detection. Their study









12

used micro-Raman optics and multichannel detection, allowing

the collection of sufficient information for both qualitative

and quantitative analysis in an on-line situation. The

detection limit for a dye, crystal violet, was determined to

be on the order of 600 molecules.

Ni et al. [33] have also used an FIA system to test the

feasibility of using colloidal-SERS for the analysis of HPLC

eluents. In this study, they observed an increase in the SERS

signal from uracil with increased temperature. They also

measured the effect of pH on the SERS signals from uracil,

cytosine, adenine and guanine. The scattering from these

bases was highest under acidic conditions. Detection limits

in the 100 nmol range were reported.

Cotton's group [34] recently used a multichannel analyzer

to detect the SERS signals from four purine bases which had

been separated by HPLC. Adenine, guanine, xanthine and

hypoxanthine were separated using reverse-phase HPLC.

Injecting increased amounts of analyte increased the width of

the peak, but not the height. Detection limits in the sub-

nanomolar range were reported.

Physical Studies

Silver hydrosols have also been used, by Heard et al.

[35], to study competitive adsorption onto a surface.

Competition was observed between pyridine and citrate for the

silver surface. It was found that chloride assisted the

pyridine in displacing the more strongly adsorbed citrate from









13

the surface. The observation of this competition using SERS

added credence to the chemical enhancement theories, since it

indicated that surface complex formation was important. The

addition of a stabilizer to the colloid before adding the

pyridine reduced the effect of the chloride.

Joo et al. [36] showed that it was possible to follow

surface induced reactions using SERS. Aromatic mercaptans,

which were chemisorbed onto the colloidal silver surface,

underwent dissociation. This reaction appeared to be a

surface induced photoreaction resulting in a scission at the

S-C bond nearest the aliphatic group. The location of the

scission is opposite the result produced via pyrolysis of the

same molecule.

Colloidal silver has also been used, by Garrell et al.

[37], as a SERS substrate in non-aqueous systems. Colloids

were prepared in a number of organic solvents including

acetonitrile and dimethylformamide. These two solvents proved

to be the most suitable for SERS analysis. The SERS spectrum

of t-butylamine on an acetonitrile silver sol was presented as

the first SERS spectrum from an anhydrous colloidal system.

Colloidal systems, such as these, may have use in non-aqueous

systems where electrochemical surfaces cannot always be used.

Several papers since the landmark paper of Creighton et

al. [23] have described the SERS signal from pyridine adsorbed

onto silver surfaces. In one study by Dawei et al. [38] a

silver chloride surface was used. The SERS activity of the









14

silver chloride colloid was seen to decrease with increased

exposure to light. The treatment of silver chloride colloids

with thiosulfate, ferricyanide, or hydrogen peroxide reduced

the SERS intensity of pyridine. Similar results were observed

for silver bromide colloids by the same group [39].

The SERS spectra of halide ions associated with the

surface were measured on a silver colloidal surface by Wetzel

et al. [40]. Chloride was found to enhance the signal of

pyridine on both gold and silver colloidal surfaces. The

addition of pyridine did not increase the intensity of the

SERS from the halide ions. The increase in the pyridine SERS

signal was attributed to the formation of a surface complex by

the pyridine which was mediated by the halide ions.

The SERS spectra of cyclohexene and pyridine on silver

colloids were measured by Yamada et al. [41]. The SERS

intensity from pyridine and cyclohexene was compared with the

intensity of the SERS active bands of water. The strong SERS

enhancement for these chemisorbed species, relative to water,

along with shifts in both the frequencies and the relative

intensities of the bands, were seen as supporting the chemical

enhancement mechanism.

Moskovits at. al. [42] observed that an increased

concentration of molecules at the surface could produce a

change in the orientation of the molecule relative to the

surface. At low coverages, 2-naphthoic acid seemed to lie

flat on silver colloidal surfaces, interacting through both









15
the w bonds of the ring system and the carboxylate group. At

higher concentrations, 2-naphthoic acid was oriented

perpendicular to the surface, adsorbed exclusively via the

carboxylate group.

The effect of pH on the adsorption of 2,2'-bipyridine

onto silver colloids has also been studied. Kim et al. [43]

observed that decreasing the pH using HBr was found to

increase the SERS signal from 2,2'-bipyridine. An opposite

trend was observed when HI and HC1 were used. Any change in

the pH caused a shift in the SERS band positions.

The SERS spectra of several amino acids and nucleotide

bases adsorbed onto colloidal silver surfaces were studied by

Suh et al. [44] to determine their orientation on the surface.

The majority of these molecules were found to lie flat on the

surface. An exception was noted for m-aminobenzoic acid,

which at low pH adopted a geometry of either standing up, or

tilting away, from the surface with adsorption occurring

exclusively via the carboxylate group.

Adenine and eight of its derivatives were investigated by

Kim et al. [45] using silver hydrosols. The effects of pH

and analyte concentration on the resulting spectra were

reported. Changes in the pH were affected by the use of

phosphoric acid. Acidic conditions were necessary for SERS to

be observed from these DNA bases, except when pyridine had

been added. This demonstrated that the addition of acid, or

some other adsorbate, might be necessary to induce aggregation









16

of the silver colloid. Of this series, only adenine was

observed to undergo a coverage dependant change in

orientation. A similar study was undertaken by the same group

for uracil and three of its derivatives [46]. Like the

adenine derivatives, these compounds preferred to lie flat on

the surface under the conditions studied. Uracil adsorbs end

on in the deprotonated form, but assumed a flat orientation

when additional sodium borohydride was added. The change in

orientation was attributed to the decrease in the surface

potential caused by the addition of the sodium borohydride.

The SERS signals from citrate adsorbed onto a slightly

aggregated silver hydrosol and a highly aggregated silver

hydrosol were compared by Blatchford et al. [47]. Both of

these silver hydrosols were protected using citrate to prevent

further aggregation. When additional sodium borohydride was

added, the SERS signal observed from the citrate decreased.

The addition of the sodium borohydride served to decrease the

potential of the surface.

The concentration and temperature dependence of the SERS

spectrum of PABA on a silver colloidal surface was studied by

Suh et al. [48]. The intensity of the 1452 cm'1 band was shown

to be linearly dependent on the PABA concentration. The

system was studied at pH 7, making the anion the predominant

species. The molecule was determined to be adsorbed flat on

the surface. A similar conclusion was reached for PABA

adsorbed onto a silver electrode surface by Park et al. [49].










SERS on ChromatoQraphic Substrates

In 1984, Tran [50] reported the first SERS spectra from

a chromatographic surface. He reported the SERS spectra of

four dyes adsorbed onto a cellulose filter paper. Limits of

detection for these dyes ranged from 0.5 ng for crystal violet

to 240 ng for methyl red. The analytes first were mixed with

silver hydrosols and then introduced onto the surface for

analysis using a syringe. A helium-neon laser was used as the

excitation source. For the case of crystal violet, the

surface enhancement was combined with a resonance Raman

enhancement (SERRS). This additional enhancement helped to

account for its much lower detection limit.

Later that year, Tran [51] performed a similar study

after first separating three dyes using paper chromatography.

These chromatograms were dried and then sprayed with a silver

hydrosol using a nebulizer. The SERRS spectrum of crystal

violet, malachite green, and basic fuchsine were evaluated on

two commercially available chromatography papers, yielding

detection limits of approximately 2 ng for these dyes.

A study by Berthod et al. [52] followed, with an

analysis of the analytical application of silver colloids to

coated filter papers. The SERS spectra of PABA and several

other nitro and amino containing heterocycles were obtained,

but no quantitative results were given. Electron micrographs

indicated that the coated filter paper surface is not

conducive to theoretical modeling, because of the entrapment








18

of silver colloids within the filter paper, as well as upon

the surface.

Several nitro containing heterocycles were detected by

Laserna et al. [53] on a silver coated filter paper, both

alone and in mixtures. This paper also demonstrated that the

components of a mixture could be spectroscopically resolved

using SERS. Detection limits in the low ppm range were

achieved with an RSD of 15%.

The first SERS spectrum from a silica gel plate was

obtained by Sequaris and Koglin [54]. They reported the SERS

spectrum of 9-methylguanine using 514.5 nm excitation. The

analyte first was mixed with the colloid and then applied to

the surface. The detection limit was determined to be 120 pg.

Sequaris and Koglin [55] later separated ten purine

derivatives using HPTLC; the plates were dried and then

sprayed with colloidal silver. Detection limits in this

latter study for these biological compounds were estimated to

be less then 5 ng/spot. Koglin [56] used a micro-Raman

instrument in a similar study to improve the spatial

resolution of this technique; detection limits were in the ng

range for purine, benzoic acid, and 1-nitropyrene, but the

spatial resolution was improved to 1 Am.

In addition to the work mentioned above, there have been

several reviews. A book on SERS, edited by Chang and Furtak

[57], was published in 1982 and offers the best introduction

to the field. A review by Seki [58] cites all references to









19
the SERS effect before 1985. Seki's review provides a good

place to start a search from, but lacks text describing the

observed results. Chang [59] has also published a report on

the status of SERS research of electrode interfaces. Cotton

[60] has recently published a review of SERS investigations

of compounds of biological interest.















CHAPTER 3

THEORETICAL FOUNDATIONS

The Raman Scattering Process

Raman scattering arises from the inelastic scattering of

photons by molecular vibrations [61]. Raman scatter is

dependent upon a change in the molecular polarizability during

a vibration, whereas a change in dipole moment is required for

infrared absorption. The Raman spectra reflect vibrational

and rotational motions, which cause changes in the molecular

polarizability. The experimental data within this

dissertation only involves vibrational transitions, therefore

only these will be discussed in this chapter and the appendix.



For a light of a frequency, vL (s '), the electric field,

E, (V/m) of the wave can be expressed as


E = Ecos(2vvLt) (3.1)



where E0, (V/m) is the amplitude of the wave. For a vibration

at frequency, vV (s ), the polarizability at any time, t, will

be



a(t) = a0(l+cos(27vvt)) (3.2)









21
where a (C2.m /J), is the molecular polarizability. Since the

electric field induces a dipole moment, A (C.m), in the

molecule through the molecular polarizability


A = -aE (3.3)


the time dependence of the induced dipole moment is given by

the supposition of the equations 3.1 and 3.2


p = aoEoCOS(2 vLt)+ (1/2)aoEoCOS(2w(viLvv)t) (3.4)


The first term corresponds to the light scattered elastically

by the molecule (Rayleigh Scatter), the second contains the

anti-Stokes and Stokes Raman (inelastic) scattering terms.

The anti-Stokes term is represented by the summation of the

frequency of the laser and the frequency of the vibration.

The Stokes term is shown as the difference between these

frequencies. The physical result is the appearance of bands

on either side separated by AvV from the exciting frequency

(vL) as shown in figure 3.1.
The oscillating electric field of the incident radiation

may, because of molecular symmetry, induce a dipole moment in

a molecule whose axis is not parallel to the applied field.































Figure 3.1. a) A schematic Raman spectrum showing the stokes
and anti-stokes Raman scattering shifted Av,, for the first
vibrational transition and, Av2, for the second vibrational
transition, relative to the Rayleigh scatter at vL. b) A
schematic diagram illustrating the origin of the Raman
scattered frequencies.



















LASER
LINE


ANTI-STOKES SHIFTED


STOKES SHIFTED


VIRTUAL LEVELS


STOKES SHIFTED


LASER LASER
LINE LINE




2
1


ANTI-STOKES SHIFTED


MOLECULAR
VIBRATIONAL LEVELS


a)


LASER
LINE









24
The relationship between the induced dipole and the applied

field is best shown by the matrix form of the polarizability

tensor



x = axx axy exz Ex

y = a a"w ayz Ey (3.5)

z. = azx a C (3.5)E


For spontaneous (normal) Raman scattering, the polarizability

matrix is real and symmetric, so the square matrix can be

diagonalized by choosing a principal axis. This attribute

allows one to calculate the magnitude and orientation of the

induced dipole.

The intensities of the Raman bands can be calculated

using the formula for the intensity of radiation, I, emitted

by an oscillating dipole.



I = (164fv /3c ')2 (3.6)


where c, is the speed of light. If the Stokes shifted lines

originate from the ground vibrational level, then the anti-

Stokes lines must originate from higher vibrational levels.

Since the populations of these vibrational levels follows a

Boltzmann distribution, the ratio of the relative intensities

of the Stokes and anti-Stokes bands can be expressed by

Is/Ias = ((VL+Vv)4/(VL-'v)4) exp(-hv,/kT) (3.7)








25

The selection rules for the harmonic oscillator allows only

Av =1, although often overtones (2vv) and combination bands

(vV, v2) are found due to anharmonicity in the molecular

vibration [62]. The presence of a surface directly adjacent

to the molecule may lower its symmetry, resulting in the

appearance of bands that are Raman forbidden in the free

molecule [61,2]. The selection rules then become "propensity

rules," with the z2 terms being the most intense. The z axis

extends outward into the surrounding medium, perpendicular to

the surface, and the x and y axes are parallel to the surface.
2 2
The xz and yz terms would be next most intense, and the x y ,

and xy terms would be the weakest.

The SERS Enhancement Process

The forces acting on a molecule located at a distance, r,

from a surface may have a variety of sources [63]. The most

important of these is the electric field of the laser. The

image field, which accounts for the polarization of the metal

by an induced dipole located near the surface, and the field

caused by the presence of induced dipoles located near the

molecule, may effect the absorption and resonance Raman

spectra of the molecule. The normal, non-resonant, Raman

intensity is only affected by the electric field of the laser.


(3.8)


Ar = ar'EL(r,vL) "exp[-i(vL+v)t]









26
where Ar (C-m), is the induced dipole, ar (C2m 2/J), is the

polarizability, EL(r,vL) (V/m), represents the field generated

by the laser at r, vL (s1 ), the incident laser frequency and,

v, (s ), the vibrational frequency of the mode.

From this equation, one can see the beginnings of the two

classifications of theories concerning the evolution of the

SERS effect. The first enhancement mechanism is termed

electromagnetic. This mechanism attributes the large

enhancements to local field enhancements due to either

resonances within the surface or points of high curvature.

This would result in a larger field term, EL(r,vL). The second

enhancement mechanism involves chemical effects; chemical

effects are due to direct interactions between the molecule

and the surface, which cause changes in the polarizability of

the molecule, ar

Electromagnetic Enhancement Theory

For the calculation of purely electromagnetic

enhancements there are three requirements. First, one must be

able to represent adequately the polarization of the molecule,

usually as a point-dipole. This condition is a "far field"

approximation, where the distance between the separation of

charge is negligible when compared with the distance between

the molecule and the surface. Second, the surface must have

a relatively simple shape, so that it may be easily modeled.

Third, Maxwell's equations must hold even at distances as

small as 1 A. The last is needed because the field felt by









27
the dipole as it nears an irregular surface becomes

increasingly inhomogeneous, causing the model to break down.

On smooth surfaces, the enhancement is observed to be

approximately 10-102 [64]. The enhancement from the smooth

surface should decay rapidly on an atomic scale as the

molecule is moved away from the surface. Spherical surfaces

are easy to model and able to couple directly with optical

fields [65].

The local field, ELc, caused by a surface located at the

center of the sphere is given by the equations



Etoc o [(E L)-Eo)/ (vL)+2c0] a3 (3.9)


where a is the diameter of the sphere, e(v) is the dielectric

function of the metal, and Eo is the dielectric function of

the surrounding medium. Each of these terms is composed of a

real, e,, and imaginary, e2, component as shown in equation

below



e = e1 + ie2 (3.10)



If the frequency of the laser is near the resonance frequency

of a sphere, the enhancement is inversely proportional to

square of the imaginary component of the dielectric function

of the metal surface, [e,] 2, at that frequency.









28
It is important to have a small imaginary component for

the dielectric constant when the resonance condition,

[l,(v)] = 2co is satisfied. Figure 3.2 shows the real and

imaginary components of the dielectric functions of bulk

copper and silver in a vacuum. Figure 3.2 can be used to

determine at which frequencies the surface plasmons of the

metal can couple with the incident radiation. The surface

states of a sphere can be in resonance with not only the

incident radiation, but also the scattered radiation. This

resonance condition allows the Raman scatter of the molecule

to be amplified [4]. Figure 3.3 schematically compares the

spontaneous (normal) Raman and SERS processes.

Prolate spheroids can be used to theoretically model

colloids which have undergone agglomeration. The

electromagnetic equations are similar to those for a sphere,

but illustrate the difference between of the local field at

the end and the sides of the spheroid. For the resonance

condition (E(VL) = -2c0), the local field at the tip of the

spheroid is enhanced by, G, the same amount as a sphere



G =I -1/e212 (3.11)


The enhancement on the side of the spheroid will be reduced

from this by a factor of E2. The tip also allows the

efficient coupling of the scattered Raman radiation with the































Figure 3.2. The real, el, (solid symbols) and imaginary, E2,
(open symbols) components of the dielectric functions of
copper (circles) and silver (triangles) [71].












































2.90


3.60


energy (eV)


S.=50
1.50


2.20


4.30


5.00













P4H 4
H r- &

$ 4
0.c i.,
00.
MS4

.g0
400
4)






4) 4)tr

0
4))

0>O4
*H ;
St
1-4 *4)




4 44m






44pC
= Q)




10 M

0 0
r- 4)
.H l 1

4 E-H14



E4 *

o wg1o




< 0- 0
-M 4





Vt 0



a 0Cl4
-U 4 ) U



*H U., 0
4-)
0 V) P
to~f








32
LU











LU
+










<^

w
LU L







Il v^



VV W^JJ^^J









33

surface, so that if vR vL then the scattered radiation

undergoes additional amplification.

If this occurs, a molecule interacting with the spheroid tip

will experience an enhancement of



G = le-1/214 (3.12)



with this being reduced by a factor of E4 for the sides of the

spheroid. This enhancement arises from the small radius of

curvature of the tip of the spheroid, not the aspect ratio of

the spheroid [66]. The basis for the increased field at the

tip of the spheroid is that in the maintenance of an

equipotential surface, the image charge would need to be close

to a point of high curvature. This translates to a more

intense field outside that portion of the surface. Figure 3.4

illustrates the variation in the enhancement factor for a

prolate spheroid as a function of polar angle.

The electromagnetic model accounts for the long range

enhancement observed by Murray et al. [67]. The results from

calculations of the distance dependence of the enhancement

from a prolate spheroid are shown in figure 3.5.

If the enhancement was purely electromagnetic, then the

Raman excitation spectrum would be dependent only on the

nature of the metal and the vibrational frequency of interest

[63]. This type of behavior has been observed on gold and













oV o
CDO0


$4 0



"0 0


I.-I

* 0 *


gont
. -
*C






4)
-,-IJ mo




-,.4
) *0
cu*2^







4'r 0 ONI
SC0
g,4



0 44
gog
g 0







040
*.4 q 0




4 000
r0


u V

0 41
.4 44 %




Np V 0 0-













0













L0)



L (D

C)




mC2L


(si!un -qJe) JoCoej .ueueouequ '*O












4; .94
O 0

Qg
.5 .

4)C
4a4


0

.1 >1


0
MO




(4-
Ck


Sto

E0 4
-0)
UCe

U)0 *






0 0






U00



404.

EaOto
0


0 .4.1 S

>9 ,
40


0 01
o 0






:5 1
Cr Q)












































(sl!un *qje) JO01oe-j luweueoueU 'D











silver surfaces for water, but not other adsorbates [59].

The observed correlation between the extinction profiles and

elastic scattering excitation profiles of non-aggregated

colloids can be accounted for by the electromagnetic theory.

The electromagnetic theory cannot account for differences in

the Raman excitation profiles for various molecules, since it

does not handle chemical effects [68]. Contrary to this is

the success of Weitz el. al. [69] in calculating the SERS

excitation spectrum of nitrobenzoate on silver island films

using the electromagnetic model.

Critics of the electromagnetic theory consistently point

out that it does not account for changes in the selection

rules and over-estimates the enhancement [70]. This tendency

to over-estimate the enhancement may be due to the use of the

bulk phase dielectric constants [65]. The use of the bulk

phase dielectric constants has been shown to be valid only for

cases where the length or diameter of the particle is less

than 1/15 of the exciting wavelength [4]. This breakdown been

observed in the measurement of the dielectric functions of

thin metal films [71].

Chemical Enhancement Theories

Chemical enhancement mechanisms require strong

interactions between the molecule and the surface. These

interactions result in an alteration of the polarizability

matrix elements of the molecule. This alteration is

attributed to either charge transfer [72,73], excitations of









39

electron hole pairs requiring "adatom" type surface defects,

or chemisorption induced resonance [74]. All of these models

are short range mechanisms, predicting no enhancement past the

first one or two monolayers adsorbed onto the surface.

This appears to be the case for an experiment involving

cyanide and pyridine [75]. The SERS bands from these

adsorbates were enhanced by up to six orders of magnitude.

The spectra were highly potential dependent, exhibiting SERS

only for the molecules that were strongly adsorbed to the

electrode surface. Competitive adsorption, similar to this,

has been observed on silver [35] and copper [24] colloids as

well.

Of these mechanisms, only the "adatom" model of Otto et

al. [76] has stood the test of time well. The charge transfer

and chemisorption induced resonance models do not specify the

necessity of active sites [4]. The enhancement from smooth

surfaces is small and can be accounted for by electromagnetic

mechanisms without the need to invoke the chemical models

[70]. The small enhancement observed on smooth surfaces is

commonly used to discount the effects of chemical enhancement

mechanisms.

The adatom model employs surface defects to break the

translational symmetry parallel to the surface. These

variations in surface morphology mediate the coupling between

photons and plasmons. This reduction in the symmetry of the

surface is also used in the electromagnetic model. The major









40

difference between the two theories on this particular point

is how far away the molecule can be away from the surface and

still feel a difference between a surface defect and the

planar surface. This is related to the relative degree of

roughness present in the surface. For the adatom model the

surface features are composed of several atoms or a defect in

the crystal structure, while the electromagnetic model uses

surface features on the order of 10-50 nm in the calculations.

Since surface defects mediate the photon plasmon

interactions, molecules adsorbed onto the surface at or near

these defects will be more easily excited. Coupling between

the electronic states of the molecule and these surface states

would account for both the breakdown of the selection rules

and the observed continuum background of inelastic scatter

[63].

Experiments in UHV conditions have examined the effect of

annealing surfaces on the SERS activity. In these

experiments, [77,78] Raman active surfaces were heated to

anneal them. Annealing the surface resulted in a loss in

Raman scattered intensity. In another experiment [59],

pyridine was adsorbed onto silver surfaces under UHV

conditions. For three low index faces, only one band

(992 cm ), attributed to physisorbed pyridine, was detected.

For a silver 540 face, the band at 992 cm1 was shifted to 1003

cm-1 due to chemisorption. In all cases this band was
cm due to chemisorption. In all cases this band was









41
unenhanced, which casts doubt on the importance of atomic

sized defects proposed in the adatom model.

There is also some data on colloids which is in conflict

with the adatom model's use of small surface defects. Cotton

et al. [34] have recently been using elevated temperatures (up

to 900 C) to increase the Raman scattering from molecules

adsorbed onto silver colloids. According to the adatom model

this heating should result in a decrease in the Raman scatter,

since much of the finer surface morphology would be smoothed

by the heating of the colloids, as annealing does to surfaces

in UHV conditions.

There is currently no theory which describes or explains

all of the SERS results. The previous section was compiled to

give a general overview of the competing theories. Under some

conditions, one theory may appear to adequately describe the

observed results. For aggregated colloidal systems, the

electromagnetic theory seems to best account for the observed

results. There are, however, indications that chemical

mechanisms do play some part. The electromagnetic model can

not account for shifts in the positions of the bands, or the

breakdown of the selection rules.














CHAPTER 4

COLLOIDS AND AGGREGATION

Metal colloids have been the focus of much attention.

The spectral properties of colloidal metals were first

quantitatively examined by Mie in 1908 [79]. Mie investigated

the scattering of light by gold particles of sizes

approximately the same size as the wavelength of light used

for excitation. His work formed the basis for further

experimental and theoretical work on what came to be known as

the Mie scattering phenomenon [80]. Absorption spectrometry

and light scattering techniques are still used to determine

the particle size and degree of polydispersity for colloidal

solutions.

The optical properties of colloidal spheres suspended in

aqueous solution can be described by the following parameters



Qext = (2/a')E (2n+l){Real(an+bn)} (4.1)

Qs = (2/a2)E (2n+l){lan2 + Ibn2} (4.2)

Qab = Qext Qsc (4.3)


where Qext is the extinction efficiency, Qsc, is the scattering

efficiency, Qb, is the absorption efficiency, a, is a size

parameter equal to (2wa/A), and an and bnrepresent a series of








43

coefficients dependent on the size of the particle and the

dielectric constant of the surface relative to that of the

medium. For very small particles, with radii less than 1/15

of the incident wavelength, these equations become



Qca = (8a4)/{31(e-1)/(e+2) 2} (4.4)
Qa = -4a{Im{(e-1)/(e+2)}} (4.5)


where Im represents the imaginary components of the dielectric

function of the metal surface relative to the surrounding

medium, and Qext is the summation of equations 4.4 and 4.5.

These simplifications are only valid for the Rayleigh limit,

when contribution from multipoles higher than the dipole can

be neglected.

The above equations have been used to calculate the

extinction spectra of spherical particles of silver and copper

in water [81,24] shown in figure 4.1. Peaks in the extinction

spectra enable one to follow the aggregation of a colloidal

system using conventional absorption spectrometry. It may be

possible to observe several peaks which correspond to several

different particle sizes [23], or only a single peak for one

particle size [24].

Another method of particle sizing is based on scattering.

Photon Correlation Spectroscopy (PCS) uses a diffusion size

model to determine particle size. Corrections can easily be

made for the refractive index of the particle at the













0

0
0
u









4,
r-l



C
1-I

0










N
0
m



*4-





0
4)
0










41
U
+>
0


4-)
U





4-,



o.
Hr














0
+-4
c4






















F-4 L-i
(0;
u *
rc)l
**p






E *-




















































O
LO 0 0
I,,- LOcJ
d d d


eoueqJosqV eA!~jeley


0
0
cO


0
LO
N


0
0
1--


0

LO
oC









o
0

0 .
0 c



0





Iq,
0
0
0



LO




m
14n
n3









46
scattering wavelength. Photon correlation spectroscopy yields

the distribution of the particle sizes, as well as, an average

particle size. One of the limitations of this method is that

the calculations usually assume a single mode distribution.

Electron microscopy can also be used to determine the range of

particle sizes, but care must be taken to avoid causing

further aggregation when preparing the sample for analysis

[82].

The aggregation process is controlled by the surface

charge of the particle and the thickness of the diffuse layer

surrounding the particle [83]. For colloidal particles, the

attraction is due mainly to van der Waals interactions. The

strength of these interactions vary as d6 with the

interparticle separation distance. The attractive forces, Ua,

(J/mol) for a spherical particle of radius a, is given by

A 8a2(d+a) d(d+4a)
U(d) --[ + 21n 1 (4.6)
12 d(d+a) (d+2a)2

where d is the distance between the surfaces, a is the radius

of the particle, and A is an empirically determined constant.

A is usually on the order of 10-19 J. From this equation, one

can expect uncharged particles to associate and eventually

precipitate from solution.

Since the majority of colloids have charged surfaces,

there are also repulsive forces to consider. The charge on

the surface is usually due to the adsorption of ions onto the

surface. There is also a diffuse layer, which shields the









47
surface charges from each other. This controls how near

similarly charged particles can approach each other. The

extension of this layer into the surrounding solution is given

by

*(d) 40(a/d) exp(a-d/X,) (4.7)

where 0 (V) is the potential of the surface and Xa (m) is a

distance parameter. The thickness of the diffuse layer is

related to the ionic strength of the solution through the

expression


Xa eeR (4.8)
F21

where e is the dielectric constant, eo is the permittivity of

free space, F (C/mol) is Faraday's constant, and I (mol/L) is

the ionic strength of the surrounding medium. From

expressions 4.7 and 4.8, we see that increasing the ionic

strength of the solution will compress the double layer,

resulting in increased aggregation. This accounts for the

"salting out" effect.

The repulsive forces, Ur, between two particles with ion

atmospheres surrounding them is difficult to calculate. The

trends observed follow the form


Ur(d) 27a3 2 ln(l-exp-dx') (4.9)
(l+a/Xa)2 ee,









48
The total potential energy of the system is the sum of

equations 4.6 and 4.9.

There are several ways to compress the diffuse layer and

induce aggregation [84]. The first is to increase the ionic

strength of the solution, so the extension of the diffuse

layer into the surrounding solution is reduced, making

collisions more probable. Another method is to adsorb neutral

compounds onto the surface, replacing the ions adsorbed onto

the surface and thereby reducing the surface charge of the

particle.

Cotton et al. [34] recently used temperature increases to

induce aggregation. This technique increases the velocities

of the particles in solution, so that the diffuse layer is not

as effective at preventing collisions between particles. More

quantitative explanations of the effect of increasing the

ionic strength of the solution, of raising the temperature and

of decreasing the surface charge of the particle on the

thickness of the double layer surrounding the particle can be

found in a treatment using the Gouy-Chapman theory [85].

The kinetic aspects of working with colloids also need to

be considered. A paper by Weitz et al. [84] describes the

aggregation process in fractal terms. The basis for the

application of fractal analysis is the self-similarity of the

colloids on various scales. The fractal dimension of a gold

colloid was shown to be 1.75. The use of fractal scaling of

the relationship between time and the mean radius of the









49

particles, yielded a logarithmic plot with a slope of 0.57 or

1/1.75. The fractal approach to quantifying colloidal

aggregation seems promising since it is accurate and provides

results which are easily visualized.















CHAPTER 5

SILVER COLLOIDS FOR SULFA DRUG ANALYSIS

Introduction

The development of drugs for therapeutic use both in

humans and animals has increased dramatically over the last

century. Over 10+8 kg of one particular class, sulfa drugs,

were administered in 1985. Of the 5000 sulfa drugs tested for

antibacterial activity, less than 30 are routinely used [86].

Sulfapyrimidines are an important subclass of these

sulfonamides (sulfa drugs). Sulfamethazine, sulfamerazine and

sulfadiazine are members of this group [87]. These drugs are

able to block the formation of dihydrofolic acid from p-

aminobenzoic acid, thereby inhibiting the biosynthesis of

folate cofactors in bacteria [88].

Sulfa drugs are not commonly prescribed to humans

anymore, because of the development of other antimicrobial

agents and the increased resistance of bacteria to these

drugs. Sulfa drugs are still widely used in veterinary

medicine, since they are easily administered. Often two or

more of these drugs are used in combination to produce the

necessary dosage. Crystalluria, the formation of crystals in

the renal tubules, can be caused by administering drugs which

have limited solubility. Since drugs of the same class do not









51

affect the solubility of each other strongly, the risk of

crystalluria is reduced.

The use of these drugs to treat livestock diseases has

resulted in the deposition of sulfa drug residues in the

tissue of swine. In 1973, the U.S. Food and Drug

Administration set a tolerance of 100 ppb for sulfonamides in

edible tissue. The current method for testing for this class

of compounds does not yield information about the identities

of the compounds found [86].

Most spectroscopic methods for determining sulfa drugs

involve diazotinization. The standard method of testing for

sulfa drugs is the Bratton and Marshall method [89], which

yields colored products by diazotinization; the colored

products can then be detected using absorption spectroscopy.

Later this procedure was modified by Iskander et al. [90] to

take advantage of the kinetics of the diazotinization to

differentiate between these compounds. This has some success

allowing differentiation between different classes of sulfa

drugs.

Diazotinization was used by Sato et al. [91] to allow

resonance Raman spectroscopy of these compounds to be measured

with excitation frequencies in the visible region. Detection

limits were estimated to be in the 10.8 M range.

Unfortunately, it was not possible to differentiate between

the various sulfa drugs spectrally.









52
Membrane electrodes were prepared by Yao et al. [92] and

used for the determination of Sulfa drugs. These membranes

were specific for sulfa drugs with detection limits in the pM

range but, were not able to differentiate between molecules of

the same class.

SERS on colloidal silver substrates will be shown to

provide qualitative and quantitative information for

sulfadiazine, sulfamethazine and sulfamerazine at levels in

the low ppb range.

Experimental

Apparatus

All of the spectra were obtained by irradiating the

samples with 100 mW of 514.5 nm line of an argon ion laser

(Spectra Physics, series 2000). The scattered radiation was

collected at 90 degrees and resolved using a 0.85 m double

monochromator (Spex Industries, model 1403) with the bandpass

set at 10 cm The detector was a cooled photomultiplier tube

(RCA, model C31034) coupled to photon counting electronics.

The acquisition, storage, and processing of the data was

carried out on an IBM compatible PC. All spectra presented

are single scans without any spectral smoothing.

Chemicals and Procedure

All chemicals were reagent grade or better and used

without further purification. Demineralized water was used

throughout the analyses. The sulfa drugs analyzed,

sulfadiazine, sulfamethazine, and sulfamerazine, were









53

purchased from Sigma Chemicals. The 100 ppm stock aqueous

solutions were prepared with 3% methanol to aid in

dissolution.

The silver colloids were prepared as follows: 100 ml of

an ice cold 1.75 X 103 M solution of AgNO3 was added dropwise,

with stirring, to 300 ml of an ice cold 7.2 X 10"4 M NaBH4

solution. An absorption spectrum of the resulting hydrosol

showed a narrow band with a maxima at 400 nm. Such a maxima

is indicative of a relatively monodisperse hydrosol containing

silver particle between 1 and 50 nm in diameter. The particle

sizes were also measured by filtration through polycarbonate

track-etched membranes, followed by SEM analysis. Each sample

analyzed was prepared by adding 0.1 ml of the analyte solution

to 0.9 ml of the room temperature colloidal solution. After

mixing, 0.5 ml of this solution was transferred to the sample

cell and analyzed.

Results and Discussion

The normal Raman and SERS spectra of sulfadiazine,

sulfamerazine and sulfamethazine are shown in figures 5.1 and

5.2 respectively. The shifts of the peaks are given in tables

5.1 and 5.2 in units of reciprocal centimeters. There are

several peaks in the normal Raman spectra, which are present

for all three of the compounds. The bands located near 1596,

1506 and 1094 cm1 in the these spectra are due to stretches

in the ring systems. The strong bands near 1150 cm1 are due

to symmetric vibration of the SO, group; these bands are









54
-1
accompanied by weaker vibrations near 1300 and 543 cm ,

attributed to the same group. The shoulders found in these

spectra at about 1635 cm'1 are commonly attributed to the amine

group. A similar feature found at 745 cm'1 is also attributed

to the amide.

Comparing these frequencies and relative intensities with

those found in the SERS spectra leads one to think that the

molecules are lying flat on the surface. The SO2 bands are

shifted from 1150 cm1 to 1114 cm"I, and are still among the

most intense in the spectrum. The other two modes appear near

550 and 1308 cm'1 for sulfamerazine and sulfamethazine, but not

for sulfadiazine. The bands near 1635 cm-' found in

sulfamethazine and sulfamerazine are shifted to 1630 cm .

This band appears in the SERS spectrum of sulfadiazine as

well at 1622 cm-1. The ring vibrations 1596, 1506, and 1094

cm are reduced in relative intensity, but not shifted

significantly. The large shifts in the vibrational

frequencies, as well as, the relative intensities of the bands

from the SO2 and amine groups indicate that they interact

strongly with the surface. From this information, the molecule

appears to lie nearly flat on the surface. This has

previously been suggested for molecules, such as DNA bases,

which also have more than one functional group.












4o0
* o


Cw

a4
14
$ 4

0
Ia

O,
#a













m u


8t.
010





to 0-4
4) I.
Me
044
' o*
01I
O N0




a)
0) (0

0 0

Id *to
aN
*-4 0)



E04






rni
0 C








q 0 r
SE C
.1d


SrCI
. P c
inu 10
0 .-
(U e Q
c r) k4
& fl) "










56







o
o



o
-.
0














I 0-'-
o








0
S4-(
o





o
i 1 o
0
C;








0
oD










0c





(0
n I) l u oo u
^ < ^ -
< \ 1 2

S S ^

^ ^ 1 r
f \ / o Q
\ I M Bo3












Table 5.1.
Raman Bands of Sulfa Drugs (in Solid Phase)
Raman Shift (cm )


sulfadiazine sulfamethazine sulfamerazine

229, sh 237, w 248,vw


286,
319,
339,


373,
402,
455,


544, w


578,
636,
661,

706,
750,
800,


vw
w
vw

vw
vw
w


273,
282,
300,
328,
342,
365,
380,

456,
504,
547,

585,
634,
659,
681,
715,
743,


vw
dbl,w
dbl,w
w
vw
w
w

w
w
m

w
w
vw
w
vw
m


290,
290,
330,


386, vw

439, w
529,vw
538, w
567, vw
579, w
637, w

685, w
716, vw


827, dbl,m 822, dbl,m 829,
848, dbl,m 840, dbl,m 842,
892, w 873,
939, vw 963, vw
995, m 999, m 999,
1098, s 1094, s 1092,
1150, vs 1153, vs 1144,
1191, w
1259, w 1241, vw
1312, vw 1298, w 1306,
1338, w 1333, w 1345,
1371, vw 1385,
1407, vw 1413, w 1416,
1438, vw 1440, vw
1466, vw
1478, vw
1505, w 1506, m 1509,
1581, sh 1560, w 1559,
1598, vs 1596, vs 1596,
1629, w 1640,
bb = broad band, sh = shoulder, m =
w = weak, v = very weak


dbl,m
dbl,m
vw

m
w
vs


vw
vw
vw
vw



w
vw
vs
w
medium, s = strong,










0) 0)
4JE
*9
W .4
. N




ua
0V

0,
,t






4*e
r*-4


.r 0,
4 )o
044
S





o i>
-H 4

S0V

to
0 *H 4
.N









0
4-4)
40
Ir4 k
|gc










r N

S43-r










* H -*
Naa
.r-4
















OH (
-40 0


0 E
1^a




a2
p^~
*

r OQ
a > i





























o
-o
{ i
i 's


o





cc
o



0
OC
o





0
o
o
o




0


O



0 0 0 0 0 0 0 0 0 00


(s,!un *qJe) Al!suelul peJalleeoS ueuwel












Table 5.2.
Surface Enhanced Raman Bands of Sulfa Drugs on Colloidal
Silver
Raman Shift (cm )



sulfadiazine sulfamerazine sulfamethazine



230, sh 230, sh 238, sh


574,
634,
684,

818,
974,


1068, sh
1114, vs
1186,vw


1340, m

1410, w


1502,
1594,
1630,

3050,
3258,
3378,


290,
342,
410,
462,
508,
544,
582,
626,
674,
762,
830,
976,
1002,
1066,
1114,
1174,
1252,
1312,
1356,
1390,

1454,
1502,
1594,
1630,
2930,
3068,
3258,
3396,


sh
vw
vw
vw
vw
vw
m
vw
w
s
m
vw
vw
sh
vs
vw
vw
m
sh
m


396, vw
458, vw
516, vw
554, w
590, m
638,vw
674, w
738, m
834, m
968, bb

1054, w
1112, vs
1186,vw
1260, vw
1302, vw

1394, m


1506,
1594,
1622,
2922,
3074,
3242,
3380,


bb = broad band, sh = shoulder, m = medium, s = strong,
w = weak, v = very


294, sh
364, vw
416,vw
458, vw
506, vw









61

From the chemical structures shown in figure 5.3, the

similarity and differences between these compounds is evident.

The SERS spectra reflect these similarities and differences as

well. The 2930 cm1 band, which is present with different

intensities for both sulfamethazine and sulfamerazine, is

absent in the sulfadiazine spectrum. The 2930 cm"' band is due

to the methyl groups on the pyrimidine ring. The locations of

the stronger peaks also differ slightly. The symmetric

stretch of the SO2 group is located at 1114 cm'I for

sulfadiazine and sulfamerazine, but is found at 1112 cm"' for

sulfamethazine. Similarly, a shoulder is seen at 1622 cm1 for

sulfamethazine, but is shifted to 1630 cm' in the other

spectra. There are also a number of peaks which appear in one

of the spectra, but not the others.

Figure 5.4 illustrates the ability of SERS to provide

quantitative information at ppb levels. The bands at 685 cm-'

and 1092 cm1 are attributed to vibrations of the benzene ring.

Variations in the intensities of these bands may be due to

slight oreintational changes of the molecule relative to the

surface. This presents a true test of the fingerprinting

capability of SERS. The intensity of the peak at 1114 cm"1 in

the SERS spectrum of sulfamethazine is useful for determining

concentrations of analyte between 10 and 500 ppb (see Table

5.3). This band was used for all of the analytical

determinations because it was linear with concentration over

several orders of magnitude. These results help to ensure


































Figure 5.3. Structures of the three sulfa drugs studied:
sulfadiazine, sulfamerazine, and sulfamethazine.











O
II
u SNH-
-- / I I K i-_I I

Sulfadiazine


O
11
SNH-
II
0


Sulfamerazine


II
u


Sulfamethazine


H\
N
Id!


N- /
H/


NM
H/

















4O
44





4J
cu



















O
,.4










0







0
0
M,Q













0
04
0











U4
*H




a4
(0

'o
0
0d










0
4iu
m












0o
0) A
C f
*i- Q





4-1



3 &










0r
































































0 0 0 0
-t C0 -


(S!iun "qje) Alisueaul peJealeoS ueweu



























Table 5.3.
Analytical Figures of Merite for Sulfa Drugs on the Silver
Colloidal System.
Compounds LOD's Sensitivity R
(ppb) (ppm )

sulfadiazine 1.0 66.74 0.997
sulfamerazine 1.0 143.82 0.999
sulfamethazine 10.0 16.62 0.999
a LOD is the limit of detection using three standard
deviations above the noise, the sensitivity is the slope of
the calibration curve, and R is the correlation coefficient of
the log-log plot.









67

that SERS will eventually provide high quality qualitative and

quantitative information as a method of routine analysis.














CHAPTER 6

A STUDY OF COPPER COLLOIDS

Introduction

The Raman scattering cross section can be increased up to

six orders of magnitude by bringing the molecule close to a

surface [6,10]. Copper, gold and silver surfaces have

generated the largest enhancements [58,69]. Because of this

enhancement, Raman spectroscopy can compete more effectively

as a trace analysis technique with other molecular

spectroscopies, such as fluorescence and phosphorescence. The

advantage of SERS over these techniques is two-fold. First,

not all molecules fluoresce or phosphoresce, whereas all

molecules have vibrational modes accessible to SERS. Second,

SERS spectra provide a wealth of information, allowing one to

identify the molecule and/or differentiate it from a similar

molecule adsorbed on the surface. These two reasons provide

the impetus for the development of SERS as a trace analytical

technique. Reviews by Garrell [1] and Campion and Woodruff

[5] have focused on the application of Raman techniques to

analytical problems.

Silver and gold colloids were first used as SERS

substrates in 1979 by Creighton et al. [23]. Much of the

research on SERS has utilized silver colloids as the









69

substrate. Silver colloids are easily produced, provide a

fresh surface for the adsorption of analyte molecules, and can

be excited by an argon ion laser. Silver hydrosols have been

used for the quantitative analysis of drugs [28,70,71] and

nitrogen containing heterocycles [72,53].

Silver hydrosols have been adapted for use with flow

injection analysis [30,31,33,29] and HPLC [34,52] systems. On

line measurements increase the reproducibility of colloid

preparation, thus lowering limits of detection. SERS has also

been adapted to both paper [52,54] and thin layer

chromatography [55] in efforts to further increase the

detection power and identification power of these separation

techniques.

Copper colloids were first evaluated for SERS in 1983 by

Creighton et al. [24]; in this study, the SERS enhancement of

pyridine by a citrate stabilized copper hydrosol was

investigated. A later study by Heard et al. [35] found there

was competitive adsorption between pyridine and citrate onto

the silver colloid surface. To minimize this interference, an

unprotected copper colloid was used in the present study. The

investigation of a copper colloid for use as an analytical

SERS substrate is reported for the first time. The colloidal

surface is prepared by a simple borohydride reduction of

cupric nitrate. These results illustrate the bright future of

SERS on copper colloids for analytical use.











Experimental

Apparatus

In Figure 6.1, a schematic diagram of the experimental

setup is shown. A Spectra Physics 2040 argon ion laser was

used to pump a dye laser (Spectra-Physics, model 375). DCM

(Exciton) was used to produce the necessary wavelengths. All

measurements were made relative to the 100 mW of 660 nm

exciting radiation. The stability of the output power and

lasing frequency were continuously monitored using both a

power meter and a 0.3 m monochrometer (McPhereson 218) with a

R928 Hamamatsu photomultiplier tube. The laser beam was

directed through the sample twice and the scattered radiation

collected at 90 degrees. A 0.22 m double monochrometer

(Spex 1680B) with slits set for 6 cm"1 resolution was used at

a scan rate of 3.0 nm/min for wavelength discrimination. The

Raman scatter was detected using a thermoelectrically cooled

(R636 Hamamatsu) photomultiplier tube. The photoelectron

pulses were amplified and counted using a (Stanford, model

440) photon counter with a 1 s acquisition time. The data was

collected on an IBM PC. The Raman bands used for quantitative

measurements were 1010, 1397, and 1387 cm1 for pyridine, p-

aminobenzoic acid, and p-nitrobenzoic acid, respectively.











W



0
C


0






0
U

r-I
u





4-1)

40



a)









a)

*-4



00
r0






*H
00


8.- ,

I0
o00






C
10
rz4







P4):





























Cfr
LUi




z
0



Z

0

Kc


) 0U
z
-7
I-
0
0


dS ---------^ OC
cr cr
(I Cc










Chemicals and Procedure

A 30 ml volume of a 2 X 10.3 M cupric nitrate solution was

added dropwise to 70 ml of a 1.0 X 10.2 M sodium borohydride

solution at room temperature with stirring. The analyte was

added as a 1.00 ml aliquot to the resulting solution. The pH

of the solution was then adjusted as needed using a small

amount of nitric acid. A sample of this pH adjusted solution

was then placed in a standard cuvette. Measurements began two

min. after the first two reagents were completely combined.

All glassware had been cleaned thoroughly with concentrated

nitric acid and the rinsed repeatedly with Barnstead water.

Pyridine, p-nitrobenzoic acid (PNBA), methanol and sodium

borohydride were obtained from Kodak (Rochester, NY). Cupric

nitrate and p-aminobenzoic acid (PABA) were obtained from

Fisher Scientific. The solutions of PABA and PNBA were

prepared in 20% (v/v) methanol. All chemicals were analytical

reagent grade or equivalent and used without further

purification.

Results and Discussion

Pyridine was chosen as the test compound for the initial

optimization because it is soluble in water, has a large Raman

cross section, adsorbs strongly to colloidal surfaces, and

SERS has been observed previously on copper colloids (figure

6.2). The enhancement factor for the pyridine 1010 cm'1 band,

corrected for prefilter and post filter effects was calculated

to be 1.8 X 105 which is in good agreement with previous









74

results [24]. Pyridine (Pk,=5.19) proved a good choice,

because of the relatively small effect that pH had upon its

SERS signal compared to those of the other analytes (figure

6.3). The optimal pH for the observation of SERS from

pyridine differs from that previously observed using a copper

electrode [18] and a different copper colloid [24]. The

optimal pH for the copper electrode and the copper colloid

used in this study differ by a 1.5 units, which may be due to

differences in the surface morphology or potential. The pH

difference between the copper colloids in the present study

and those used by Creighton et al. [24] is about 2.5 units.

This is probably due to their use of citrate to stabilize the

colloid. The adsorbed citrate will affect both the surface

potential of the colloid and the adsorption of the pyridine

onto the surface. The difference in the pH dependence of the

SERS intensities of PNBA (Pk,=3.45) and PABA (Pk,=4.65, 9.30)

are due to the differences between their dissociation

constants. The acidic form of the amino group on PABA will

adsorb more strongly than the basic form. At slightly basic

pH's, PABA is bound to the surface via both functional groups.

The effect of pH on PNBA is probably due mainly to the surface

charge of the colloid. The narrow pH ranges within which PABA

and PNBA exhibit SERS presents an additional method for differ

entiating between similar compounds.

The absorption spectrum of the colloid over time (figure

6.3) shows the growth of a peak at 570 nm. It has been












3^ .
* 0 V
C ftlP
*. 4 o
v 0 0
4 14


0 U
00



0
to-t



00



SO
t *
H0
olo


1t4



r)



05


O H





40 0
o 0
UM









rc.
a)







.o4
C
S0








04 4
30




|i4 ) (









76




o
0


0














U)

E





(um
C















(s:!un "qje) Al!suelul peielleoS uewel
% < } ^

-j ? S
-===^- -===3 < c/











W 04)

>4


( U
(0 'e
id 0
9)44




Og
(U0


Oo

0 0 O'






0
-00



10 0
944
0 >t
(a r-4 40







0






oogo
0


ot




0
c o
c1 C

















90 0 OD

0 > 0
u 4.






(U -*l4


-.-I to >4
44 tV C0'




























< 4
L










E--


ALISN3INI


Z
o 0
T- -

_J
0
(J)




0.
O_





co






N-


3AIV-1.D3









79
previously calculated that there is a correlation between the

570 nm peak and copper spheres having a diameter of z 20 nm.

The particle size distribution of the colloids was analyzed

with a Malvern 4700 photon correlation spectrometer. The data

analysis was performed using the optical constants for bulk

copper [71]. During the time span of interest, the average

particle diameter was found to be approximately 100 nm (see

figure 6.4).

The optical density in figure 6.3 reaches a maximum 780

seconds after the reagents were combined. If the 20 nm

particles are the precursors of the most SERS active particle

size, the SERS signals should increase while the 20 nm

particles are present and then decrease. This correlates well

with the SERS signals shown in figure 6.5., which is largest

for the 1397 cm'1 band at 800 s and decreases thereafter. The

time dependance of the 1397 cm1 SERS signal makes it difficult

to acquire a good spectrum using a scanning instrument

(Figures 6.7, 6.8).

The copper colloid system has a much lower background

than silver colloidal systems. Unfortunately, the SERS

signals are also smaller, so that the signal to noise ratio is

not improved. This may be due to lower adsorption onto the

copper surface as compared to silver surfaces. An additional

problem is that the copper colloid cannot be prepared more

than 30 min. ahead of time, where the silver colloid can be

made stable for future use [23,33,34].












. 0
0Q) I
*g







t o





0 .
'e -





-4J
r II
H ,





04)
4g *
0P
QrJo




U Q





?A .1 0

0

294 '
4) Q

0m0*


g o


M) 00) 0

$4 *4 )-4 0
0 4 N

4, 0 r. 0 0Q
rN -l -H I
or rl-r U 9






















































































eoueqjosqv


SDC:


Ln N d d d dC
.4 Ci d d 5


o 0c
a
D











* 0


FA
ton
a -d


tw
m *








0
*C 0


4)
Ow'r
000


Q'*


t4 O *4
V 4)4
0U1
ol 9



044





$40
.63
0O 0 g<
.0 4
U 4ffl
















$4-
0
04'1 4j






OU

0 .0











VA 44




$4 0
&'~ Q) ^
*rlX (^ ^ &I












83

















Ln
+ + + + +



+ + + + +




U-)


u-
+ + + + + "
0



+ + + + j-
+






+ + + + +



+ + + + +
\o u




+ + + + +







+ + + + +





U ) J -(\


uoilsodwodW luaejed






























Figure 6.6. Intensities of the 1606 cm" (filled triangle) and
1397 cm (filled circle) SERS bands of PABA as a function of
time. The .conditions were pH=9.0, [Cu ] = 6.0 X 10 [BH4 ]
= 1.9 X 10 and 50 ppm para-aminobenzoic acid.




















1397 cm-1


1606 cm-1


400
400


800


1200


1600


Time (s)


1 -




0-
0
O










*1


,-1 *
0oo






H*


o
0.


- 0







0 .


0o
1x4
A *



0 ,.
0
44 il






0


a,
uo



0
0
aV





4J
r/l
Wa
Wa










10
S-U
10 W




4a 0



0
rq
*ii ^
fcXUr-




























































o 0 0 0 0 o$
S00 o0 0
(sj!un "qje) Al!suehul pejeileoS ueweH=











-s

00
oa
r-l4

0
,4


$4
O
0




*d
0
o
Ca



*-,4







Ou



00


to
O




















ri8
4J.4






c 0












4m
0
I4O












So
o a
U* 4
4- 0
00




Q)
00








Co
0U


00
^^




9-3
b\g
H a
















































(SI!un "qje) A !suelul peJealeoS uejuey

































Figure 6.9. Calibration curves for pyridine, p-aminobenzoic
acid, and p-nitrobenzoic acid on the copper colloid.









































20 40 60


Concentration (ppm)


80


100