Citation
Analysis of glass, ceramic, and soil samples using laser ablation inductively coupled plasma mass spectrometry

Material Information

Title:
Analysis of glass, ceramic, and soil samples using laser ablation inductively coupled plasma mass spectrometry
Creator:
Baker, Scott A., 1971-
Publication Date:
Language:
English
Physical Description:
viii, 159 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Average linear density ( jstor )
Calibration ( jstor )
Ceramic materials ( jstor )
Laser ablation ( jstor )
Lasers ( jstor )
Particle size classes ( jstor )
Plasmas ( jstor )
Signals ( jstor )
Soil samples ( jstor )
Soils ( jstor )
Chemistry thesis, Ph. D ( lcsh )
Chemistry, Analytic -- Quantitative ( lcsh )
Dissertations, Academic -- Chemistry -- UF ( lcsh )
Laser ablation ( lcsh )
Mass spectrometry ( lcsh )
Plasma spectroscopy ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1998.
Bibliography:
Includes bibliographical references (leaves 153-158).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Scott A. Baker.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
029218554 ( ALEPH )
41372459 ( OCLC )

Downloads

This item has the following downloads:


Full Text











ANALYSIS OF GLASS, CERAMIC, AND SOIL SAMPLES USING
LASER ABLATION INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY












By

SCOTT A. BAKER


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


1998














ACKNOWLEDGMENTS

I would like to thank my research advisor, Dr. James D. Winefordner, for his

patience, guidance, encouragement, and enthusiasm. I consider myself extremely

fortunate to have had the opportunity to work under his direction. I would also like to

thank Dr. Benjamin W. Smith for many useful discussions concerning this research and

for always having an open door. Both of these people were integral in my development

as a scientist.

I am grateful to all of the members of the Winefordner research group for their

friendship. They were a source of inspiration and helped me a great deal in producing

this work. I would especially like to acknowledge those who contributed directly to my

experiments. These people included Igor Gornushkin, David Rusak, Bryan Castle, Robin

Russell, Melody Bi, Rebecca Litteral, Matthew Dellavecchia, and Ricardo Aucelio.

Additionally, I would like to acknowledge Dr. Kobus Visser, Dr. Oleg Matveev, and Dr.

Nico Omenetto for many useful discussions during the course of my graduate studies.

I thank Jeanne Karably for answering many questions and for making sure things

went smoothly. Also, I would like to acknowledge Steve Miles for his electronics

expertise, and Joe Shalosky, Gary Harding, and Dailey Burch of the machine shop for

helping me on many occasions.

I am grateful to Dr. Paul Holloway and Dr. Steve Pearton for use of their

profilometers, and to the staff at the Major Analytical Instrumentation Center for








allowing me to use their facilities. They also produced the electron microprobe data.

My parents have always provided me with love and support. They have been a

great source of strength and encouragement throughout my life. I cannot express enough

how much the love, support, and friendship of my wife, Amy, means to me each and

every day. She has always been there to keep me going, and to keep things in

perspective. I look forward to our spending a long and joyous life together.

Finally, I would like to acknowledge the National Science Foundation-

Engineering Research Center for Particle Science and Technology and the Air Force

Office of Scientific Research for funding this research.














TABLE OF CONTENTS


ACKN OW LED GEM EN TS................................................................................................ ii

ABSTRA CT......................................................................................................................vii

CHAPTERS

1 IN TRODU CTION ................................................................................................... 1

2 BA CKGROUND ..................................................................................................... 6

Inductively Coupled Plasm as....................................................................... 6
Inductively Coupled Plasma-Mass Spectrometer Interface....................... 10
Ion Optics............................................... ................................................... 12
M ass Analyzer........................................................................................... 13
Detection of Ions........................................................................................ 14
Sample Introduction in ICP-M S................................................................ 14
Laser Ablation............................................................................................ 16

3 GENERAL CONSIDERATIONS AND OPTIMIZATION
OF EXPERIM ENTAL PARAM ETER S............................................................... 19

Introduction ............................................................................................... 19
Instrum entation.......................................................................................... 19
Sam ples...................................................................................................... 21
Results........................................................................................................ 22
Plasm a Operating Conditions........................................................ 22
Scan Parameters and Internal Standardization............................... 33
Sampling Strategy.......................................................................... 35

4 AN ALY SIS OF GLA SS SAM PLES..................................................................... 41

Introduction................................................................................................ 41
Experim ental.............................................................................................. 42
Calibration Procedure................................................................................ 43
Results....................................................................................................... 47
Fractionation.................................................................................. 47
Solution-Based Calibration....................................................... 53








Spot Sam pling V ersus Line Sam pling........................................... 58
Conclusions................................................................................................ 69

5 ANALYSIS OF SILICON NITRIDE CERAMIC BEARINGS............................ 70

Introduction................................................................................................ 70
Experim ental.............................................................................................. 72
Results........................................................................................................ 73
Identification and Distribution of Elements in Silicon Nitride......73
Results from Solution- and Glass-Based Calibration.................... 78
Measurement of Mass Ablated and
Estim ation of System Effi ciency....................................... 82
Conclusions................................................................................................ 84

6 INVESTIGATION OF LIGHT SCATTERING
FOR MASS NORMALIZATION IN LA-ICP-MS............................................... 85

Introduction................................................................................................ 85
Experim ental.............................................................................................. 86
Results........................................................................................................ 90
Scatter Signal Normalization and Comparison
w ith Internal Standardization............................................. 90
Comparison of Scatter Normalization for Glass,
Soil, and M acor Ceram ic................................................. 96
Conclusions.............................................................................................. 103

7 ANALYSIS OF SOIL AND SEDIMENT SAMPLES........................................ 104

Introduction.............................................................................................. 104
Experim ental............................................................................................ 105
Results...................................................................................................... 108
Spot Sam pling Versus Line Sam pling......................................... 108
Solution-Based Calibration.......................................................... 110
Speciation Effects........................................................................ 117
Effect of Organic Content............................................................ 121
Analysis of Particle Size Fractions.............................................. 125
Analysis of Soils U sing Standard Additions................................ 130
D election Lim its........................................................................... 134
Conclusions.............................................................................................. 134

8 EVALUATION OF A COMPACT LASER SOURCE....................................... 136

Introduction.............................................................................................. 136
Experim ental............................................................................................ 137
Results...................................................................................................... 138
Conclusions.............................................................................................. 146



V









9 CON CLU SION S AND FU TURE W ORK .......................................................... 147

Conclusions.............................................................................................. 147
Future W ork............................................................................................. 150

REFEREN CES................................................................................................................ 153

BIO GRAPHICAL SKETCH ........................................................................................... 159










































vi














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

ANALYSIS OF GLASS, CERAMIC, AND SOIL SAMPLES
USING LASER ABLATION INDUCTIVELY COUPLED MASS SPECTROMETRY

By

Scott A. Baker

May 1998

Chairman: James D. Winefordner
Major Department: Chemistry

Laser ablation (LA) as a direct solid sampling method for inductively coupled

plasma mass spectrometry (ICP-MS) has been used for the analysis of glass, ceramic, and

soil samples. The strength of this technique is that it can be applied to essentially any

solid material and eliminates the need for difficult and time consuming dissolution

procedures. The major limitation of LA-ICP-MS is the presence of matrix effects,

making quantitation difficult in the absence of standards of identical composition.

This work was largely concerned with the development of methodology for

obtaining accurate quantitative results from a variety of solid materials, including glasses,

ceramics, and soils without the need for matrix-matched standards. The use of solution-

based calibration for the analysis of a National Institute of Standards and Technology

(NIST) glass resulted in excellent agreement (typically +/- 10%) with the certified values.

Detection limits were in the sub-ppm range for all elements studied. The solution








calibration technique was also applied to silicon nitride ceramic bearings and the

accuracy of the results were confirmed with X-ray microanalysis. In addition, standard

glass materials were useful in the analysis of ceramics provided that differences in

ablated mass were properly accounted for. To this end, the use of light scattering for

measuring the amount of ablated material was evaluated. Light scattering was effective

for mass normalization provided that the particle sizes of the ablation products were

sufficiently similar.

The use of solution-based calibration for soil samples resulted in poorer

agreement than in the case of glasses and ceramics; however, results were still within +/-

20 % for most analytes studied. Studies involving sample particle size, organic content,

and element speciation were performed to understand the effects that these variables have

on LA-ICP-MS measurements. The use of standard additions was briefly studied. It was

concluded that this technique is useful, provided the particle sizes in the sample are

sufficiently small.

A compact, inexpensive laser was also evaluated for LA-ICP-MS measurements.

It was used for the analysis of glass and aluminum samples and provided low to sub-ppm

detection limits for the analytes studied.













CHAPTER 1
INTRODUCTION

Developments in materials research, environmental science, industrial quality

control protocols, and numerous other endeavors have placed a high premium on the

development of accurate and sensitive analytical techniques. The presence of trace

element impurities in ceramics, for example, have been shown to affect the physical and

mechanical properties of these materials (1). In the electronics industry, the presence of

alkali and alkaline earth metals may cause corrosion and degradation of microelectronic

devices. In addition, the alpha emitters U and Th can cause errors in storage devices (2).

Therefore, the presence of these elements and their concentrations must be accurately

determined. Geologists need to determine absolute and relative concentrations of

elements in a variety of geological samples, in an effort to discover the principles

governing their distribution and migration (3). In addition, the ability to precisely

measure isotopic ratios of such elements as Li, Sr, Nd, Pb, and U is important for the

radiogenic dating of materials and providing an understanding of terrestrial processes (3,

4). The effects of heavy metals (e.g. Pb, Cr, Hg, and Cd) on human health are well

established and techniques for monitoring suspected areas of contamination, such as

soils, are required (5, 6). Furthermore, research has shown that the presence of particular

elements (e.g. Se) results in some protection against heavy metal toxic effects; therefore,

it is important to be able to monitor the presence of several elements in the sample of








interest (7). It is evident that precise and accurate measurement of multiple elements at

low levels is required in a variety of applications.

In many cases the sample of interest is a solid and the ability to analyze the

material directly would be beneficial. Direct analysis of solids eliminates the need for the

time-consuming digestion and dissolution procedures required for many materials. In

addition, there is a reduced risk of sample contamination, incomplete digestion, or the

loss of some analytes during sample preparation. Lower absolute detection limits are also

possible since there is no dilution of the sample prior to analysis. In some instances, the

local concentration of analytes is of greater importance than the bulk concentration and

this information is completely lost with dissolution of the original sample. In these cases,

the ability to probe particular areas of the solid is essential and techniques capable of

microanalysis are needed.

Several of the techniques commonly used for the elemental analysis of solids are

X-ray fluorescence (XRF) (8), electron probe microanalysis (EPMA) (9), secondary ion

mass spectrometry (SIMS) (10), and glow discharge mass spectrometry (GDMS) (11).

Each of these techniques has its strengths and weaknesses. XRF is a well established

technique for the analysis of conducting and nonconducting solids; however, it suffers

from relatively poor detection limits (10-100 ppm) and significant matrix effects when

analyzing thick samples (3, 8). In addition, it is essentially a bulk technique and provides

poor spatial resolution. EPMA is a valuable surface analysis technique due to the high

spatial resolution (0.1-0.5 tm) that it provides. The technique is limited by high relative

detection limits (100-1000 ppm), sample charging with nonconducting samples, and the

requirement that the sample be held in vacuum. In addition, quantitative work requires a








highly polished surface and strict matrix-matching (9). SIMS is another technique for

surface analysis, but unlike EPMA, it offers very high sensitivity with detection limits as

low as 1 ppb (10). It is widely used for depth profiling studies because of its high in-

depth resolution (1-2 nm). Lateral resolution is typically around 0.1-1 Igm (10). Like

EPMA, sample charging is a major problem when analyzing nonconducting samples and

the sample must be held in vacuum. In addition, matrix effects are very severe making

quantitation extremely difficult (12, 13).

GDMS is a technique which has been widely used for the analysis of conducting

samples. This is due to low detection limits (1-100 ppb) and high measurement precision

(11). Nonconducting samples can be analyzed by diluting the sample in a conductive

matrix (11), using a secondary cathode (14), or by operating the glow discharge in the rf

mode (15). The latter technique has shown considerable promise for the analysis of

nonconductors (e.g. ceramics and glasses), but a major challenge to accurate quantitative

measurements is matrix effects based on sample thickness. The GD technique can be

used for depth profiling studies with a resolution approaching 100 nm. The lateral

resolution; however, is usually only a few millimeters (16). A major benefit associated

with GDMS, like SIMS, is that it provides isotopic information by virtue of its detection

methodology.

The use of laser ablation inductively coupled plasma mass spectrometry (LA-ICP-

MS) for the analysis of solid samples compares favorably with other techniques. A

focused, pulsed laser beam can be used to sample essentially any type of material and

when combined with the high sensitivity of ICP-MS, the technique offers sub-ppm

detection limits for most elements. Because the sampling and excitation steps are








separate, each can be independently controlled for optimum performance. This results in

reduced matrix effects compared to single step laser based techniques, such as laser

microprobe mass analysis (LAMMA) (17). Laser sampling also allows for modest

spatial profiling of the sample with lateral resolutions on the order of microns and in-

depth resolution on the sub-micron scale. Mass spectrometric detection also allows for

the measurement of isotopic ratios. The versatility of the technique is evidenced by the

many types of samples which have been analyzed, including metals, glasses, ceramics,

geological materials, biological materials, and polymers. One major advantage of LA-

ICP-MS compared to other solids techniques is that the sample can generally be analyzed

without any modification and is held at atmospheric pressure. This results in rapid

sample changing and increased throughput. In addition, elemental sensitivity factors for

LA-ICP-MS are relatively uniform for most elements when compared with many other

solids techniques. This makes the technique excellent for semiquantitative analyses (12);

however, quantitative measurements by LA-ICP-MS are often difficult to obtain.

The major difficulty associated with laser sampling is the complex nature of laser-

material interactions (18). This results in an unknown quantity of material being ablated

with a composition which may or may not accurately reflect the bulk sample. The

properties of the ablated material (e.g. composition and particle size distribution), depend

on the laser and material characteristics. Because of the complex nature of the laser

sampling process, matrix-match standards are often required for quantitative

measurements on a given sample. This represents the major obstacle to the widespread

applicability of the technique, since for many solid matrices, suitable standards are not

readily available. To overcome these difficulties, approaches such as the use of fused








glass beads (19, 20) or pelletized mixtures of diluent and analyte (20, 21) have been

explored. Although useful in some cases, the sample preparation that is required

eliminates a major advantage associated with laser sampling and any spatial variations

cannot be assessed.

The intent of the present work is to develop methodology for the accurate and

precise determination of analytes in a variety of solid matrices, including glass, silicon

nitride ceramic, and soil, by LA-ICP-MS. Optimum experimental parameters were

determined and the effectiveness of solution-based calibration for the analysis of solids

was evaluated. In addition, the utility of glass standard reference materials for calibration

of ceramics was examined. As an alternative to internal standardization, mass

normalization by measuring light scattering from the ablated material was studied. Its

applicability for various sample types was assessed. A compact, inexpensive laser source

was evaluated for the analysis of glass, copper, and aluminum samples. Future research

directions are suggested based on these studies.













CHAPTER 2
BACKGROUND

The inductively coupled plasma (ICP) has developed into the dominant excitation

source for elemental analysis. Its initial development was as an excitation source for

atomic emission spectroscopy (AES); however, in the past decade it has been widely used

as an ion source for mass spectrometry. Numerous publications dealing with the

fundamentals of inductively coupled plasma-mass spectrometry (ICP-MS) and some

applications are available (22-27). The major benefits of ICP-MS, compared to ICP-

AES, are superior detection limits, spectral simplicity, and the ability to measure isotope

ratios. Detection limits for solutions are typically 100-1000 times better for ICP-MS,

with typical values for most elements in the 1-100 ppt range. Mass spectra of the

elements are simple and unique. The natural isotopic abundance spectral pattern provides

a simple means of identifying sample constituents. The ability to measure isotopic ratios

is important in geological and nuclear applications; however, it also allows one to assess

the existence and extent of interference. The major components of a typical ICP-MS are

the ICP, an interface system, ion lenses, mass analyzer, and a detector. These will be

discussed in greater detail.

Inductively Coupled Plasmas

The ICP is a partially ionized gas maintained by energy coupled to it from a radio

frequency (RF) generator, typically operated at 27 or 40 MHz. Energy, usually between

1 and 2 kW, is coupled to the plasma with a 2-4 turn copper coil which acts as the








primary of an RF transformer. The plasma itself acts as the secondary. Argon (Ar) is

typically used as the plasma gas, although hydrogen-argon, nitrogen-argon, xenon-argon,

and helium plasmas have been studied as well (22). The plasma is generated inside and

at the end of a quartz torch, which consists of an assembly of quartz tubes. An

illustration of the ICP torch is given in Figure 2-1. The outer gas flow (coolant) is

typically 10-15 L min' and serves both to protect the tube walls from melting and as the

main plasma support gas. The middle annular flow (auxiliary or intermediate) is used to

keep the plasma from melting the central injector tube. Typical values for this flow are

0.5-1.0 L min"1. The inner gas flow (nebulizer or carrier) is used to puncture the plasma

and inject aerosol from the sample introduction system. This flow is typically in the

range of 0.7-1.0 L min"'.

The copper load coil is positioned around the end of the torch and supplied with

RF current. This produces a time varying magnetic field (27 or 40 MHz) which lies

along the torch axis. A Tesla coil is used to seed the Ar flow with free electrons. These

electrons process around the magnetic field in circular orbits and the energy supplied to

the coil is converted into kinetic energy of the electrons. At atmospheric pressure, the

free electron path before colliding with Ar atoms is only around 10-3 mm; therefore, the

plasma is rapidly heated and a bright discharge is formed (23).

At the RF frequencies used, the skin effect is occurring which ensures that most

of the energy is coupled to the induction region of the plasma. The nebulizer gas punches

a channel through the center of the plasma and there is little mixing with the outer

annular region. Heat is transferred to this central gas flow mainly by radiation and

conduction from the annular induction region. Gas kinetic temperatures in the central















Sample Aerosol
and Nebulizer G


.00


Intermediate I
Gas Coolant Gas


Figure 2-1. Inductively coupled plasma torch.








region are typically between 5000-7000 K, while the temperature in the outer region is as

high as 10,000 K (23). It is important that the power is coupled mainly into the outer

region where there is little interaction with the sample aerosol. This is because the

sample composition can vary significantly and have only minimal effects on the plasma

sustaining processes. Separation of the energy coupling and sample excitation region is

one of the major reasons why the ICP is characterized by relatively few chemical and

physical interference.


The role of the plasma is to convert the sample into free ions. Most analyte ion

formation results from collisions of electrons, and the probability of this process is

dependent on the electron density (_1015 cm"3) and temperature (26). The gas

temperature, which describes the kinetic energy of Ar atoms, controls the desolvation and

vaporization of the sample aerosol. Transit times through the center of the plasma are

typically several milliseconds; therefore, desolvation, vaporization, atomization, and

ionization processes occur very efficiently. The degree of ionization is dependent on the

ionization conditions in the plasma, which are dominated by the major plasma constituent

(typically Ar, H, 0, and electrons) and the partition function and ionization energy for the

atom of interest (23). The Saha equation, though not strictly valid since thermal

equilibrium does not exist in the plasma, is often used to give insight into ionization

conditions in the plasma. Under normal operating conditions, elements with ionization

energies of 8 eV or less (approximately half of the elements in the periodic table) are >90

% ionized in the plasma, while Ar with an ionization energy of 15.75 eV is only about

1% ionized (27). In addition, only a few elements (e.g. Ba, Sr, and Pr) have second








ionization energies low enough to produce a significant population of doubly charged

ions. The end result is a very efficient ion source for primarily singly charged ions.


Inductively Coupled Plasma-Mass Spectrometer Interface


The ability to extract ions representatively from the plasma is obviously quite

critical in ICP-MS. A description of the ICP-MS interface is provided with particular

emphasis on the interface used in this work (Figure 2-2). It consists of three cones

through which the pressure drops from atmosphere to approximately 10-5 torr by means

of differential pumping. The first cone, the sample cone, is located approximately 13 mm

from the top of the load coil and has an orifice of 1.1 mm. The region behind the sample

cone is maintained at 2-3 torr by a mechanical pump. This results in the formation of a

supersonic jet expansion. The jet consists of a freely expanding region, the zone of

silence, surrounded by shock waves called the barrel shock and Mach disc. Shock waves

result from collisions between the supersonic jet and residual gas in the expansion

chamber. To avoid losses of ions due to collisions and scattering, the second cone

(skimmer) is positioned before the Mach disc inside the zone of silence. Studies have

indicated that maximum ion transmission occurs when the sampler-skimmer distance is

approximately 2/3 of the distance to the onset of the Mach disc (28). The location of the

skimmer cone is 8 mm from the sample cone and the diameter orifice is 0.8 mm to allow

the centerline of the expansion to pass while removing the cooler gas at the edge of the

supersonic jet (29). Only about 1% of the gas that is sampled passes through the

skimmer cone. The pressure in the region behind the skimmer cone is maintained at 10-3

torr by a turbomolecular pump.











Sample Cone


Skimmer Cone


Accelerator Cone


Ion Beam Crossover


Barrel Shock Zone of Silence


Figure 2-2. Schematic of ICP-MS interface.









Unique to the Finnigan MAT SOLA used in this work is the incorporation of an

accelerator cone. It is located 10 mm behind the skimmer cone and has a +2 kV potential

applied to it. The lens is used to focus the ions that pass through it and accelerate them to

the ion optics. The purpose for this will be presented more clearly in the discussion on

ion optics.


Ion Optics


The purpose of the ion optics is to supply the quadrupole with ions of low enough

energy that they can be efficiently differentiated with respect to their mass to charge

ratios. Ions must be separated from neutrals, and photons produced from the plasma must

be removed since these can activate the detector and contribute significantly to the

background. Typically, a Bessel box has been employed for separating these

components. The Bessel Box consists of a central photon stop for removing photons,

while neutrals are removed by pumping of the chamber. Ions are electrostatically steered

around the photon stop and refocused into an exit slit. This arrangement results in two

detrimental effects; space charge effects and mass dependent focusing of the ion beam

(29). Space charge effects are more significant for lighter ions since they are more easily

deflected from their flight path. Mass dependent focusing results because different mass

ions have different kinetic energies and thus have different paths through the ion lenses.

Therefore, optimization of ion transmission is mass dependent and a compromise setting

must be used. This is typically in the mid mass range for multielement work, resulting in

a transmission profile which falls off at the low and high mass ends. The Finnigan MAT








SOLA ICP-MS accelerator cone and ion optics are designed to reduce these effects and

produce a more even transmission profile.


The accelerator cone supplies a high velocity stream of ions condensed in a

tightly focused beam. This beam is deflected off axis and realigned to exit parallel to the

original ion transmission, since ion losses of 50-80% have been measured when photon

stops are used (23). In addition, because the residence time in the ion optics is shorter

with the accelerated beam, space charge effects will be less significant. Since the beam

has not been defocused and refocused, as is the case with the Bessel Box, no mass

dependent transmission effects have been introduced. The final stage of the ion optics is

a phase matching lens that reduces ion velocities to the level required by the quadrupole.


Mass Analyzer


Several types of mass analyzers have been employed for ICP-MS, including

quadrupoles, magnetic sectors (30), time of flight (31), and ion traps (32). Only a

discussion ofquadrupoles is provided here, since it was used in this work.


The quadrupole consists of four straight metal rods arranged parallel to and

equidistant to the ion axis. Opposite rods are connected together with DC and RF

voltages applied to each pair. The DC voltage is positive for one pair and negative for

the other. The RF voltages on each pair are 180 out of phase. Appropriate amplitudes

of these potentials produce trajectories for ions along the quadrupole which are stable for

only one particular mass to charge ratio. By keeping these potentials constant, single ion

monitoring is accomplished. To produce a mass scan, the amplitudes are varied while

keeping the RF to DC ratio constant. The mass transmitted is linearly related to the








magnitude of the RF and DC voltages. Peak jumping between masses of interest is

accomplished by rapidly selecting discrete values of RF and DC voltages. The use of

rapid and repetitive peak jumping is extremely important when dealing with transient or

highly fluctuating signals, such as those typical of LA-ICP-MS. The mass resolution is

determined by the ratio of RF and DC voltages. This is typically set to provide 1 amu

peaks across the mass range (5-240 amu).


Detection of Ions


Detection of ions is most often accomplished with a channeltron electron

multiplier. This detector is characterized by high gain (~108), low dark current, and fast

response time. Output pulses from the detector are sensed by a fast pre-amplifier and

sent to a discriminator and counting circuit. The detector is used for signal rates between

1 and 106 counts/second. For higher ion currents, a Faraday collector plate is used. This

detector can be used for ion signal rates between -105 and 5 x 1010 counts/second;

however, the time constant of the DC amplifier limits the Faraday to relatively low scan

rates (64 ms or greater). In the Finnigan MAT SOLA, selection of the detector is

accomplished electronically by deflection of the ion beam to the appropriate detector.


Sample Introduction in ICP-MS


Samples are introduced to the ICP-MS as gas, vapor or a fine aerosol of solution,

or as solid particles. Pneumatic nebulization of solutions is the most common means of

sample introduction, due to its simplicity, sufficient stability, and an overall greater

understanding of the process. Although this dissertation is concerned with the








introduction of solid particles by laser ablation, mention is made of solution introduction

since it was often used in this work.


The principle behind pneumatic nebulizers is the disruption of a liquid stream by

a high velocity gas, resulting in the production of aerosol droplets. The liquid stream is

provided by a peristaltic pump which typically produces a 1 mL min' flow. In the

concentric type of nebulizer used in this work, liquid is introduced to a fine central

capillary and when the liquid reaches the tip, it is broken into fine droplets by the

shearing action of Ar gas flowing around the orifice at 0.5-1.0 L min-1. The liquid

particles produced can be as large as 100 Atm; therefore, they must be further processed

before entering the ICP-MS. This is accomplished with a Scott-type double pass spray

chamber, which acts as a low pass filter. Larger particles settle out and only particles <10

pAm are transported to the plasma.


The introduction of solutions, typically aqueous with small amounts of acid,

results in a large number of spectroscopic interference. The dissociation of water

produces large amounts of oxygen and thus the presence of oxide species can be

significant. In addition, the constituents of acids used will also contribute to the

background spectrum. For this reason, nitric acid is often the preferred acid since its

constituents (H, N, and 0) are already present in the plasma (23). Numerous methods

have been employed to decrease the solvent plasma load, such as cooled spray chambers

(33), Peltier effect coolers (34), membrane interfaces (35), and heater condensers (36).


Several techniques have been employed for introducing solids into the ICP-MS.

The uses and relative merits of these techniques are discussed in several reviews (22, 23,








27). They include slurry nebulization, electrothermal vaporization (ETV), direct sample

insertion (DSI), arc and spark ablation, and laser ablation (LA). Of these techniques, the

last (LA) is most promising since it can be used to sample a wide range of materials. In

addition to bulk analyses, it also allows for microanalysis by virtue of the focusing

characteristics of lasers. These attributes put laser ablation in a unique position among

solid sampling techniques for ICP-MS. The first application of LA-ICP-MS was reported

by Alan Gray in 1985 (37). Since that time, the technique has been used to analyze a

wide variety of materials, including metals (38, 39), glasses (19, 20, 40), ceramics (41,

42), geological materials (43, 44), biological materials (45), and polymers (46).


Laser Ablation

The ablation process (Figure 2-3) depends on a large number of factors, including

the nature of the solid material, laser characteristics (wavelength, pulse duration, etc.),

and pressure and composition of the gas medium. Nonetheless, a phenomenological

description can be given. When a short duration pulse of sufficient energy (typically

irradiances > 108 W cm -2) strikes the sample, the surface is instantaneously heated past

its vaporization temperature through linear one-photon absorption, multi-photon

absorption, dielectric breakdown, and additional undefined mechanisms (47, 48). Energy

dissipation of the surface layer is slow relative to the laser pulse duration and before this

layer can vaporize underlying layers have reached their vaporization temperature. The

underlying layers reach the critical point (temperature and pressure) resulting in an

explosion of the surface. This process is described as nonthermal and so no fractional

vaporization should occur (18). During the ablation process, a plasma is initiated on the

surface with temperatures in the range of 1041 05 K (47, 48) and a duration of several
























Atoms, Ions, Molecules, Pulsed Laser Beam
Clusters, Particles

0 0

: .. ..-.. .



S" "'...-''"' Target 'i

Target


Figure 2-3. Conceptual drawing of laser ablation.









microseconds. This also interacts with the sample surface and contributes to the quantity

and composition of the ablated material.


In addition to LA-ICP-MS, numerous other uses of laser ablation have been

explored. An excellent review by Darke and Tyson deals with fundamental aspects of

laser ablation and its uses in analytical spectrometry (17). These include techniques

which directly use the laser induced plasma as an excitation source (laser induced

breakdown spectroscopy LIBS) (49), ionization source (laser microprobe mass

spectroscopy LAMMA) (50), or atom reservoir (laser ablation laser excited atomic

fluorescence spectroscopy LA-LEAFS) (51). As in LA-ICP-MS, laser ablation has

been used as a method of sample introduction for numerous analytical techniques

including flame and furnace atomic absorption (52, 53), glow discharge atomic emission

(54), and microwave induced plasma atomic emission (55). The major advantage of

separating the laser sampling step from spectroscopic detection is that it allows for

independent optimization of each step, potentially resulting in improved analytical

performance.













CHAPTER 3
GENERAL CONSIDERATIONS AND OPTIMIZATION
OF EXPERIMENTAL PARAMETERS

Introduction

The purpose of this chapter is to outline several of the key variables that affect

LA-ICP-MS measurements. It is intended to serve as a guide in the selection of

appropriate experimental conditions. The measurements have been performed on a

variety of samples (metals, glasses, ceramics, and soils); however, the goal is to provide

information that is generally applicable to all matrices. More specific information on

particular materials can be found in subsequent chapters.

Instrumentation

An illustration of the LA-ICP-MS setup is shown in Figure 3-1. It consists of a

Finnigan MAT SOLA ICP-MS (Hemel Hempstead, UK) and System 266 laser ablation

accessory. The ablation module consists of a Spectron SL 401 Nd:YAG laser which has

been frequency quadrupled to produce an output beam of 266 nm, as well as beam

steering and focusing optics and a CCD camera for remote viewing of the sample. The

CCD camera is mounted in parallel with the laser and aids in focusing of the laser beam.

The system also includes an x-y-z translation stage for adjusting the focus of the laser

beam and for selecting the area of the sample for analysis. The system is designed to

provide analysis only at a fixed location. However, it was modified to allow for

translation of the sample while the laser was firing. Laser repetition rates of up to 5 Hz









Sample Cone


Solution -"f


Figure 3-1. LA-ICP-MS System


Ar








can be utilized, with pulse energies of- 0.1-1 mJ and pulse widths around 15 ns in

duration. When focused to a 50 4tm spot, laser irradiances (W/cm2) of- 8x1 07-8x1O08 are

achieved.

For analyses, the sample is placed in an ablation cell consisting of a 6 cm glass

tube with a quartz window for transmission of the UV laser beam. The cell has a total

volume of ~ 100 cm3. Ablated material is swept out of the ablation cell and to the ICP-

MS through 1.5 m of 3/16" i.d. plastic tubing with a flow of argon. In addition, the

output from a concentric nebulizer and cooled Scott-type spray chamber can also be

continuously introduced to the ICP-MS during ablation with a glass Y-connector. This

dual sample introduction approach is termed "wet plasma" operation (38). A comparison

of wet and dry (ablation chamber flow only) plasmas will be made later in this chapter.

The ICP is typically operated between 1100 and 1300 W with coolant and intermediate

flows of 15 L/min and 0.9 L/min, respectively. Sample is carried to the plasma, whether

from the ablation chamber, nebulizer, or a combination of both, with a total argon flow of

-1 L/min.

Samples

Various samples were used in these studies. They included National Institute of

Standards and Technology (NIST) (Gaithersburg, MD) glasses, silicon nitride ceramic

bearings, various metals, and NIST soil samples. The soil samples were pressed into 1

cm pellets at a pressure of 5 MPa. No binder was required to produce relatively rigid

pellets. The only constraints on sample size were that they fit into the ablation chamber

and the height match the range of the stepper motor used to adjust the focusing lens

position. This stepper motor could be adjusted over a range of 1.25 cm and the lower








focusing position corresponded to a point -0.2 cm above the base of the ablation cell.

Therefore, samples with thicknesses ranging from 0.2 cm to 1.45 cm could be analyzed

without any modification.

Results

Several experimental variables were investigated. These include plasma

operating conditions (wet and dry plasmas, RF power, and argon flow rate), laser

sampling mode (stationary or translated sample), and detection considerations (scan

parameters and internal standardization).

Plasma Operating Conditions

Among the attributes often cited for laser ablation as a sample introduction

technique is the elimination of solvents and hence, the resulting diminished level of many

background peaks. A comparison of background spectra from wet and dry plasmas

illustrated this point (Table 3-1). The dry plasma was operated at 950 W with a 1.0

L/min flow from the ablation chamber, while the wet plasma was operated at 1300 W

with a nebulizer flow rate of 0.65 L/min and an additional 0.35 L/min from the ablation

chamber. These operating conditions were chosen because they provided the highest

analyte signals for laser ablation measurements. In the wet plasma mode, a 2 % HNO3

solution was introduced to the nebulizer at 1 mL/min with a peristaltic pump (Gilson

Minipulse 3, Middleton, WI). Table 3-1 lists some of the major background peaks. In

each case, the background level is significantly larger for the wet plasma. In addition, the

introduction of solution can introduce trace levels of impurities even for high purity

reagents. The peak at m/z 58 was partially due to the Ni sampling cone.




















Table 3-1. Comparison of wet and dry plasma background.


Background Species
(m/z)

CO+, N2+ (28)

N2H+ (29)

NO+ (30)

ArH2 (42)

ArN+ (54)

ArOt (56)

ArO, Ni (58)


Dry Plasma Signal
(counts/s)

3.3x105

1.9x105

3.8xl106

3.2xl 04

2.5x105

1.3x105

2300


Wet Plasma Signal
(counts/s)

1.Oxl7

1.3x106

3.0x108

2.4x106

1.5x107

6.5x106

6200








The effect of RF power on analyte signal for both dry (a) and wet (b) plasmas is

shown in Figure 3-2. The 63Cu signal was measured from laser ablated brass samples as

a function of RF power. As previously indicated, the wet plasma optimizes at higher RF

powers for the same total flow (1.0 L/min) through the torch injector tube. The reason

for this behavior was the energy required in vaporizing the solution aerosol. It should

also be mentioned that RF power optimization only applied to particular flow conditions,

since these variables were interactive. For example, if the flow was kept at 1 L/min and

the ratio of nebulizer gas to ablation chamber gas was increased, a higher RF power was

required for maximum analyte signals due to the increased solvent loading. For both wet

and dry plasmas, increasing the total flow through the injector tube required higher RF

powers for efficient ionization.

Optimized plasma conditions were more easily determined for the dry plasma

since only two variables (RF power and argon flow) were involved. The RF power was

adjustable over the range of 950-1450 W for stable plasma operation. With the wet

plasma, the nebulizer flow rate was more difficult to optimize. To gain insight into the

proper nebulizer and ablation chamber flow rates, the RF power was held constant at

1350 W while the nebulizer and ablation chamber flow rates were systematically varied.

Total argon flow rates between 0.8 and 1.2 L/min with individual flows from 0.2 and 1.0

L/min were studied. These studies indicated that ablation chamber flows of 0.3-0.4

L/min and nebulizer flows of 0.6-0.7 L/min were most suitable in terms of both

sensitivity and precision.


















(a)


2.5x1i'-


2.0xidl


~11.5x1cp-


950 100 1050 1100 1150 1200 I I
950 1000 1050 I1100 1 150 1200


RF Power (W)


(b)


1100 1150 120 1250 1300 13 50
1100 1150 1200 1250 1300 1350


RF Power (W)


Figure 3-2. Effect of RF power on Cu signal for (a) dry and (b) wet plasmas.


1.4xl07"


1.2xl07-


l.Oxx107-


8.OxlO'-


6.0x10'-








Although background levels were higher, wet plasma operation was preferred for

most LA-ICP-MS measurements. This was because it provided a convenient means for

tuning the ion optics, gave better sensitivity and precision, and was useful in calibration

studies as discussed in Chapters 4, 5, and 7. Tuning of the ion optics in the dry plasma

mode involved continuously ablating a sample and adjusting the ion lenses to maximize

the signal from a matrix element. For multielement work, a mid-mass element was

chosen for tuning. Because the amount of material ablated by each laser shot varied and

different size particles were being introduced to the ICP-MS, laser ablation signals tended

to be noisier than those originating from nebulized solutions. The highly fluctuating

signal made tuning of the ion optics difficult.

With the wet plasma, solution-based analytes could be used for tuning. This

made tuning much easier since the signal was more stable. The major requirement for the

applicability of solution-based tuning was that analytes in the solution aerosol behaved

similar to those in laser ablated solid particles. Because of the high temperature of the

ICP and the constant plasma conditions (i.e. solution was continuously introduced

whether ablating or not), this appeared to be the case (Figure 3-3). These plots

demonstrated that analyte signals from laser ablated NIST 611 glass (Co, Ni, Cu, Sr, and

Ba) and a nebulized 20 ppb solution (Mg, Al, Mn, Co, Ni, Cu, Zn, Sr, and Ba) optimized

at similar RF powers. In addition, all analytes optimized at the same RF power. Similar

behavior was observed for soils, ceramics, and metallic alloys. This behavior was critical

for the use of solution-based calibration for the analysis of solids, as will be discussed in

Chapter 4.

















(a) 1.0-


0.8-


0.6-


o 0.4-
z


1000 1050 1100 1150 1200 1250 1300 1350 1400
Rf Power (W)


(b)


I


1 '-tI I


1000 1050 1100 1150 1200 1250 1300 1350 1400
Rf Power (W)


Figure 3-3. Effect of RF power on analyte signals for (a) solution and (b) laser ablated glass.


1.0-


Al 0.8-


- 0.6-


0 0.4-
0.2z
0.2-


v.-





















Table 3-2. Comparison of sensitivity and precision for wet and dry plasmas.
NIST 611 glass


Mn Ni Cu Zn Sr

Sensitivity-Dry
(cps/ppm) 26 130 49 10 41
Sensitivity-Wet
(cps/ppm) 3180 1050 1590 170 1660
PrecisionW-Dry
(%rsd) 4.8% 16% 9.4% 13% 13%
Precisiona-Wet
(%rsd) 4.9% 8.9% 7.1% 5.6% 5.9%


a For precision measurements, analyte signals were normalized by the signal from Co.








The sensitivity and precision of LA-ICP-MS measurements was different for wet

and dry plasmas. Table 3-2 compares the sensitivity (signal/ppm) and precision (% rsd)

of laser ablation measurements for several elements in NIST 611 glass with both plasma

operating modes. The concentration ofanalytes was around 500 ppm in the glass sample.

The results indicated that much higher sensitivity was obtained with a wet plasma.

Several workers have studied the effects of water in the ICP and found that the

presence of some amount of water provided higher excitation than the complete absence

of water (56, 57). This is related to the presence of hydrogen produced from the

dissociation of water. The thermal conductivity of hydrogen is about ten times that of

argon, resulting in greater energy transfer in the ICP (56). There is obviously an

optimum amount of water that should be added to the plasma. The quantity of water

introduced to the plasma could be controlled by adjusting the temperature of the spray

chamber. When the spray chamber was cooled, water condensed on the walls, resulting

in smaller quantities of water being introduced to the plasma. The effect of spray

chamber temperature on analyte signals was studied and the results for Ba are shown in

Figure 3-4a. An increase in ion signals with decreasing spray chamber temperatures was

observed for all analytes studied (Be, Mg, Co, In, Ba, Pb, and U). In addition, the

decreased solvent loading resulted in lower oxide levels as shown for ArO (Figure 3-4b).

The level of oxides for Ba and U were also studied and showed a decrease of about a

factor of two between 25 C and 0 C. Unfortunately, some background species (Ar2 and

ArN) actually increased slightly with decreasing spray chamber temperatures.













(a)


0 5 10 15 20 25


(b) 3
I

I
31


I I
0 5


Spray Chamber Temperature (C)


Spray Chamber Temperature (C)


Figure 3-4. Effect of spray chamber temperature on (a) Ba and (b) ArO signals.


1.Oxl0"-


&O8.0x105"


6.0x105-


4.0x10-


i








As indicated in Table 3-2, the precision of LA-ICP-MS measurements improved

with wet plasma operation. The reason for poorer precision in dry plasma measurements

was not entirely clear; however, it could be related to the different ablation chamber flow

rates. The lower flow rates associated with wet plasma conditions allowed for more

mixing of the ablation products from successive laser shots and produced more stable

signals. For dry plasma, less mixing occurred and signal fluctuations were greater in

magnitude. Because the quadrupole MS is a scanning instrument and spends a finite time

at each mass, highly fluctuating signals can lead to reduced precision when measuring

several isotopes. This will be illustrated more clearly in the consideration of scan times

later in this chapter.

Another potential benefit of the lower ablation chamber flow associated with wet

plasma operation was that it probably introduced smaller particles to the ICP. The

fraction of material transported through the transfer tubing as a function of flow was

estimated (Figure 3-5). The simple model (58) used in these calculations was based on

uncharged spherical silica particles and assumed that gravitational settling (rather than

convective diffusion) was the dominant loss mechanism. Arrowsmith and Hughes

studied the entrainment and transport of laser ablation products produced from Mo metal

(59). They determined that -90 % of the products were entrained, -40 % were

transported to the ICP, and concluded that the major loss mechanism was gravitational

settling. Because of differences in flow rates, tubing length, and ablation material,

different values would be expected for the present work. As demonstrated in Figure 3-5,

a significantly different particle size distribution would be transported to the ICP. For

example, 5 p.m particles would not be transported to the ICP in the low flow case, but


































1.0



0.8



0.6



0.4


1.0 L/min


\ 0.4 L/min
^
\
^
%.
*s
^>


I I I II I
0 1 2 3 4 5

Particle Size (im)









Figure 3-5. Calculated transport efficiency for silica particles at

0.4 L/min. and 1.0 L/min. Ar flow rates.








nearly 50 % of these particles would be transported with the higher flow rates associated

with dry plasma operation. This is especially important considering that studies have

indicated that particles >3 pgm are not completely vaporized and excited in the ICP (60,

61). Thepresence of these particles could lead to selective removal of more volatile

species in the ICP, resulting in inaccurate measurement of bulk concentrations. The

particle sizes produced from laser ablation are dependent on numerous factors, including

laser pulse duration, wavelength, irradiance, repetition rate, and physical properties of the

material. Studies have indicated that a wide range of particle sizes (tens of nanometers to

tens of micrometers) are produced during the ablation process (19, 21, 59).

Based on the comparison of wet and dry plasmas, all subsequent work was

performed in the wet plasma mode. Although the increased level of some interference

was a drawback, the overall gain in analytical performance made it preferable.

Interferences were usually not a problem at most masses; however where they were,

another isotope of the element of interest was usually available. For example, 57Fe could

be measured instead of 56Fe, although with a loss in sensitivity due to differences in

natural abundance.

Scan Parameters and Internal Standardization

The scan parameters (number of channels, dwell time, and number of passes) and

data acquisition mode determined the total scan time for a measurement and played an

important role in the accuracy and precision of LA-ICP-MS measurements. The number

of channels refers to the number of discrete masses measured by the detector. In this

system, a total of 4096 channels are available. Dwell time refers to the amount of time

spent at each of these discrete locations (channels). The number of passes is simply the








number of times that the region or regions of interest are scanned over and averaged

during a measurement.

A procedure which combined the relative merits of both peak jumping and

scanning was used in the majority of this work. It involved scanning over a selected

region of interest (1 amu/peak) and then rapidly jumping to the next region of interest for

scanning, and so on. This allowed for the acquisition of true peaks, and therefore time

was spent only on regions of interest. Both peak height and peak area measurements

were used. Precision was similar for both types of measurements; but as expected, peak

area measurement produced lower detection limits.

The total measurement time for laser ablation analyses was typically around one

minute (300 shots @ 5 Hz). This was chosen to provide relatively rapid analyses, but

with enough laser shots and mass ablated to introduce a representative portion of the

sample to the ICP-MS. The amount of material ablated was dependent on the laser

irradiance and sample type, but was typically in the range of 2-50 ng per shot. It was

important to determine appropriate scan parameters for providing accurate and precise

measurements.

For these studies, NIST 611 glass was ablated and nine different isotopes (43Ca,

"55Mn, 59Co, 6Ni, 63Cu, 88Sr, '07Ag, 138Ba, and 165Ho) were measured on the multiplier

detector. For each isotope, 16 channels were measured with dwell times of 2, 8, 16, 32,

and 64 ms. The number of passes (128, 32, 16, 8, and 4) was adjusted so that the same

amount of time spent at each mass (-4 s). All measurements were normalized with an

internal standard (43Ca) to account for differences in mass ablated during and between

measurements. The average % rsd for the measurements was 3.8 % (2 ms), 3.5 % (8 ms),








5.8 % (16 ms), 7.4 % (32 ms), and 8.4 % (64 ms). The latter dwell time represented the

minimum dwell time that could be used with the faraday detector. These results clearly

indicated the benefit of using short dwell times and a large number of passes to average

out fluctuations in the laser ablation signal. Dwell times of 2 to 4 ms were used in all

subsequent multielement laser ablation measurements.

The use of an internal standard (43Ca in the above measurements) was important

for providing reasonable levels of precision. This was because a variable amount of

material was ablated due to differences in laser power (20 % rsd for n = 15 laser shots),

the nonlinearity of the ablation process (18), and the changing sample morphology (crater

formation). To account for these variations, the signal from a matrix element was

commonly measured. An important assumption was that the analytes and internal

standard were distributed the same over the sampled area, and that they exhibited similar

transport to, and excitation in the ICP. Internal standardization measurements typically

improved the precision by a factor of two or so. This improvement was greater in some

cases, for example, with soils and particulate samples. In these cases, the absolute signals

would routinely vary by a factor of three or more. In addition, when analyzing multiple

samples, it was difficult to ensure identical focusing of the laser beam with respect to the

sample surface. Internal standardization provided a convenient means of mass

normalization and eliminated the need to precisely control the position of the sample with

respect to the focused laser beam.

Sampling Strategy

Initial work with the LA-ICP-MS system was confined to single spot

measurements. The temporal profile of the 21Si signal from a glass sample obtained by

















































0 20 40


Time (s)












Figure 3-6. Temporal profile of Si signal from glass obtained
with repetitive pulsing at a single spot.


4x108 1-


3x108 9-


2xO108 -


1x108 -


I I I


100 120








repetitively firing the laser at a fixed location is shown in Figure 3-6. The benefits of

repetitive pulsing, as opposed to single shot measurements, was that it allowed material

from successive ablations to mix. This provided a continuous laser ablation signal. The

MS could be repetitively scanned and information on the bulk sample was obtained.

Typically, the initial signal was higher and then fell off to a lower "steady-state" level

after 20 or 30 s. For this reason, a scan delay of 30 s was utilized in single spot

measurements to minimize any bias resulting from measuring signals in the initial region

where they changed very significantly, since the MS spent a finite time at each mass.

To determine whether the drop in signal was related to signal suppression or

simply differences in the ablation efficiency, integrated measurements of smaller

numbers of shots were performed. The laser was fired 50 times and the integrated 28Si

was measured. This was repeated in sets of 50 up to a total of 300 shots. Figure 3-7

indicates that the large initial signal was the result of a larger mass ablated rather than

signal suppression. In addition, the laser was focused both above and below the laser

surface to see if this had any effect on the observed behavior. Regardless of the laser

focus, a larger mass was initially ablated and then a relatively constant amount was

produced. This was not surprising since the first laser shots were incident on a flat

surface; whereas, all subsequent shots were directed at the crater formed in the sample.

To take advantage of the higher initial ablation yield, the instrumentation was

modified to allow for translation of the sample at 15 .tm/s while the laser was repetitively

fired. Faster translation rates tended to produce more erratic signals and also required

that the direction of the stage be changed more often because of the limited sample width.

As expected, translation resulted in improved sensitivity because of the higher ablation






























A 0.5 numm above surface








S0.5 mm below surface
* at surface


6


SA A


2


g


10


Set # (Each Represents 50 Shots)


Figure 3-7. Si signal as a function of shot number.


5x10 9 -1


4x10 9 -


3x10 9 -


2x10 9 -


lxIO 9 -








rate. The effect was greatest with pressed powder samples since the craters formed in the

surface were deeper, resulting in relatively rapid defocusing of the laser beam. A

comparison of craters produced from single spot and translation sampling is shown in

Figure 3-8. Measurements of the ablation depth and volume for these craters will be

discussed in Chapter 5. Another potential benefit of translation is that a larger area on the

surface can be sampled, minimizing the effects of local lateral heterogeneity. A

drawback is that a smaller surface layer is removed when translating and surface

contamination and/or heterogeneity can be problematic. This will be demonstrated with

glass samples in Chapter 4.







(a)















(b)
















Figure 3-8. Laser produced craters in silicon nitride with
(a) single spot and (b) translation sampling.













CHAPTER 4
ANALYSIS OF GLASS SAMPLES

Introduction

Numerous studies on the analysis of glasses by laser ablation inductively coupled

plasma mass spectrometry (LA-ICP-MS) can be found in the literature. These studies

deal with a wide range of applications, from elemental fingerprinting of crime scene

evidence (62) to the use of glasses as calibration standards for geological materials (63-

66). In the glass industry, as the number of formulations and applications of glass

materials continues to grow there is an increased demand for rapid, accurate, and precise

determinations of the concentrations of elemental constituents (67). Direct methods of

analysis are preferred because of the difficulty in digesting these materials. Dissolution

generally consists of three steps: (1) treatment with HF; (2) further oxidation by addition

ofHNO3, HC104, or H202; and (3) final addition of HCl or HN03 (68). When digesting

in an open vessel, silica is lost by formation of volatile SiF4. An alternative to wet

chemistry is melting the sample with a suitable flux, such as sodium hydroxide or lithium

metaborate. The major problem with this method is the production of high salt

concentrations, which can lead to significant matrix effects and/or block the nebulizer

and ICP-MS sample cone (68). Because of the difficulties and time required for both

methods, LA-ICP-MS is an attractive alternative; LA eliminates the need for extensive

sample preparation and also offers the potential for information on the spatial distribution

of elemental constituents.








Glasses have been studied in this work for several reasons. They provided a

relatively homogeneous, analyte-rich matrix for characterization of the LA-ICP-MS

system. They were certified for several elements and therefore provided immediate

feedback on the suitability of non-matrix matched calibration strategies. As previously

mentioned, this was of particular interest because of the lack of matrix-matched standards

for many materials of interest (e.g. ceramics). The use of solution-based calibration for

the analysis of glasses will be addressed in this chapter and some of the important

findings discussed.

Experimental

The LA-ICP-MS system has been described previously (Chapter 3). Table 4-1

lists the typical ICP-MS operating conditions. The laser was operated at 5 Hz with

energies ranging from 0.1-0.7 mJ. Both single spot and line sampling were used.

NIST glass samples (611, 612, 614, and 617) were used in these studies. These

are synthetic Si, Na, Al, and Ca glasses which have been spiked with 61 different

elements at nominal concentrations of 500 ppm (NIST 611), 50 ppm (NIST 612), 1 ppm

(NIST 614), and 0.02 ppm (NIST 617). They are in the form of 1-3 mm thick discs.

For the solution-based calibration studies, a 10 ppm multielement standard (High

Purity Standards Charleston, SC) was diluted with deionized water and Optima-grade

HNO3 (Fisher Scientific, St. Louis, MO) was added to bring all solutions to 2 % HNO3.

Concentrations of the standards ranged from 1 to 50 ppb. A 2 % HNO3 blank was

continually introduced during laser ablation analyses via a glass Y-connector at the base

of the ICP torch. In this way, identical plasma conditions were maintained whether an

ablated solid or nebulized solution was being introduced.












Table 4-1. ICP-MS operating conditions.


Rf Power

Coolant gas flow rate

Auxiliary gas flow rate

Nebulizer gas flow rate

Ablation chamber flow rate

Solution uptake rate


Scan Conditions

Faraday Scans (major elements)

Scan range per isotope

Number of passes

Number of channels per amu

Dwell time

Multiplier Scans (minor elements)

Scan range per isotope

Number of passes

Number of channels per amu

Dwell time


1200 W

15 L/min

0.9 L/min

0.6 L/min

0.4 L/min

1.0 mL/min







1 amu

32

8

64 ms



1 amu

128

16

4 ms








The ion lenses, nebulizer gas, and RF power were optimized using the (1 "In)

signal from a 1 ppm solution (measured on Faraday). Both the Faraday and multiplier

detectors tuned at almost identical ion lens settings; therefore, either detector could be

used with the optimized settings. An ICP-MS response curve (Figure 4-1) was generated

for typical tuning conditions using a multielement solution. The absolute sensitivity

(number of ions detected / number of atoms introduced) was calculated for nine different

analytes (Be, Mg, Ni, Co, In, Ce, Pb, Bi, and U) which encompassed the mass range from

9 amu (Be) to 238 amu (U). The resulting plot indicated that the instrument was most

sensitive for mid-mass analytes and fell off for low and high mass analytes. It was

investigated whether tuning with a low (24Mg) or high mass isotope (208Pb) had any effect

on the instrumental response curve. No significant differences were observed.

Typical scan parameters are listed in Table 4-1. For most multielement analyses,

a total scan time of 60-80 s was used. This resulted in the averaging of 300-400 laser

shots per measurement. A 30 s delay was typically used before the start of data

acquisition as discussed in Chapter 3. Both peak height and peak area measurements

were used. Peak heights were measured directly using the SOLA instrument control

software. Peak area measurements were performed by importing the ASCII scan files

into a spreadsheet program for integration of the spectra. Typically, 8 channels were

integrated for each peak. Temporal profiles ofanalyte signals were generated by

sequentially and repetitively monitoring masses of interest. This feature was

incorporated into the SOLA software suite; however, due to problems in the program,

only a limited number of masses could be monitored.
































Ni, Co


Bi
IK


50


100


Mass


ICP-MS response curve generated by nebulization
of a 1 ppm multielement solution. Sensitivity refers
to the number of ions detected per number of atoms
introduced to the ICP-MS.


4.0xO10 -5 -





3.0xO -5 -





2.0xlO -5





10x10 -





0.0


Figure 4-1.








Calibration Procedure

The feasibility of using solution standards for the semiquantitive analysis of solid

materials has been studied previously (38, 67, 69-71). Hager (71) developed a model that

used response factors determined from solution nebulization and modified them based on

element-dependent volatilization efficiencies. This work reported accuracies of only +/-

50 % for aluminum, steel, and copper standards. The technique does not appear to be a

general method because ablation of nonmetals is likely to produce molecular species,

such as oxides and silicates. In addition, a large fraction of the ablated material

transported to the ICP is solid particles that have undergone solid-liquid-solid, rather than

solid-monatomic gas, phase changes (69). The use of dual-sample introduction appears

to be a more promising alternative for analyzing materials where suitable calibration

materials are not available. Its utility has been demonstrated for geological samples (38,

67, 70) and metals (69).

The calibration procedure used in this work involved measuring the intensity of a

single isotope for each element of interest (including an internal standard) in solution

standards and determining relative sensitivity factors (RSF's). The RSF is defined in

equation (4-1), where E represents the element of interest and IS represents the internal

standard.

(4-1) RSF(E IS) = Sensitivity(E) = Intensity(E) Concentration(IS)
Sensitivity(IS) Concentration(E) Intensity(IS)

If(a), identical plasma and mass spectrometer conditions prevail when nebulized

calibration solutions or laser ablated solid particulates are being introduced to the ICP-

MS (see Figure 3-2) and (b), the ablation material is an accurate representation of the

bulk material, then RSF's obtained from solutions and solids will be equal, i. e.








(4-2) RSF(E/IS)soiution = RSF(E/IS)soIid.

Provided that the concentration of the internal standard is known in the solid of interest,

the RSF's can be used to calculate the concentration ofanalytes from the expression

(4-3) Concentration = Intensity(E),oId Concentration(IS)oid
(4-3) Concentration(E)sola Inest(S, 1 SSIhf
Intensity(IS)Loid RSFoIuo,

A major requirement of this solution-based methodology is that the concentration

of some element (the internal standard) must be known prior to analysis. This can either

be measured by some complementary technique, such as x-ray microanalysis (EPMA), or

determined from the stoichiometry of the solid. This is obviously a limitation of the

technique and a potentially useful method for overcoming this limitation is discussed in

Chapter 6. Typically, a minor isotope of a major matrix constituent is chosen as the

internal standard. In the analysis of these certified glass materials, 43Ca was found to be

the most suitable internal standard because of relatively low levels of interference at this

mass, and because signals could be measured with the multiplier detector. Signals

greater than 106 counts/s (cps) resulted in saturation of the electron multiplier detector,

which was exclusively used for the multielement analysis of trace elements in these

glasses.

Results

Fractionation

The production of ablated material that is an exact representation of the original

sample is essential in non-matrix-matched calibration of bulk materials. For

inhomogeneous samples, this requires that enough material is removed so that local

concentration variations in the solid are averaged out. For samples that are

homogeneously distributed on the scale of sampling, the major problem is fractionation








of species during the ablation process. Internal standardization can generally account for

differences in mass ablated both within and between matrices; however, fractional

ablation is more of a problem since fractionation causes enhancements or reductions in

relative analyte signals. Several workers have studied the role and extent of fractionation

in laser ablation measurements (69, 72-78). These investigations have revealed the

importance of laser wavelength, irradiance, and sample characteristics in determining the

extent of ablative fractionation. In addition, transport fractionation has been revealed as a

source of fractionation as well (75).

Two mechanisms have been proposed for ablative fractionation: volatilization of

non-refractory elements from sample areas outside the bulk ablation area; and zone

refinement, where elements within the molten zone surrounding the ablation crater

undergo selective migration into the laser ablation spot. Recent work has indicated that

the latter mechanism is the most likely source of fractionation, since the extent of

fractionation correlated with melting points of the element compounds in the sample (75,

78). If the former mechanism was dominant, the boiling point or vapor pressure of the

element compound would be critical.

The effect of laser irradiance and wavelength on elemental fractionation has been

studied as well (69, 78). Cromwell and Arrowsmith (69) reported that the Pb/Cu ratio

produced from a brass sample with UV laser ablation was one order of magnitude higher

with lower laser irradiances (< 1.3 x 108 W/cm2) than at higher irradiances (> 109

W/cm2). They attributed the increased fractionation at lower irradiances to the increased

volume of molten sample relative to the bulk ablation volume. Recently, Figg and Kahr

(78) studied the effect of laser irradiance, as well as laser wavelength, on ablative








fractionation in glass samples. They reported much higher fractionation in the case of

ablation with 1064 nm and 532 nm wavelengths, compared to 266 nm. This agrees with

earlier studies that compared ablation behavior for IR and UV lasers (76, 77).

In the case of UV lasers, the primary means of material removal is by direct laser

interaction with the surface for the duration of the laser pulse. The initiation of the laser

induced plasma occurs in less than 1 ns; however, the UV beam is not significantly

absorbed and the role of the plasma in material removal is insignificant (77). With an IR

laser, the direct interaction only lasts a fraction of the laser pulse duration since the

plasma is highly absorbing at this wavelength. The plasma is heated by the laser energy

and the primary means of material removal is by the plasma-material interaction (77).

This produces local heating of the sample, resulting in greater fractionation of more

volatile species.

In this work, the effect of laser irradiance on relative analyte signals was studied

to determine if fractionation was occurring, and if so, to what extent. This was essential

if the goal of accurate quantitative measurements was to be realized. In this study, NIST

611 glass was repetitively ablated at 5 Hz with laser pulse energies of- 0.5 mJ. The

sample was not moved, since this would maximize local heating of the sample and induce

fractionation. The focus of the laser was adjusted over a range of 2 mm to adjust the

irradiance at the sample surface. The laser spot size, as determined by the surface craters,

ranged from 40 pgm to 200 gxm. Signals for eleven elements (1 B, "55Mn, 59Co, 85Rb, 88Sr,

89Y, 90Zr, 133Cs, 182W, 20Pb, and 29Bi) were acquired over 500 laser shots. The ratios

of intensities to the "55Mn intensity were determined to see if any changes were observed

over the range of irradiances. Mn was chosen because it's oxide melting point (oxides








should be dominant species in glass) was intermediate for those elements studied. Table

4-2 lists the oxide melting points for several elements in the glass sample.

The results for these measurements are given in Figure 4-2. The intensity ratios

have been normalized to their value at the highest laser irradiance (- 5x108 W/cm2).

Therefore, the relative change in composition can be determined by the deviation from a

value of one. In Figure 4-2 (a), elements with oxide melting points lower than Mn are

plotted and in (b), elements with higher oxide melting points are plotted. Based on these

results, significant differences in the composition of the ablated material were not

observed for the low melting point elements until very low irradiance values (< 5x107

W/cm2). Below this value, fractionation of more volatile elements occurred (as much as

2 times in the case ofBi). Since an irradiance of 5xl07 W/cm2 was near the ablation

threshold, visual observation of the laser-induced plasma could be used as a guide to

ensure that the laser irradiance exceeded this value. The presence of a visible laser spark

on the sample surface could then be used as a guide to ensure representative sampling of

these elements in glass.

The behavior of those elements with higher oxide melting points was more of a

problem. The relative intensity of Co did not deviate significantly over the range of

irradiances studied; however, this was not surprising considering the proximity of the

oxide melting points for Co and Mn. The relative intensities of Y and Zr decreased for

irradiances below -1.5xl08 W/cm2, indicating that these more refractory oxides were not

being ablated representatively. For some unknown reason, this trend was reversed at

lower laser irradiances. Nonetheless, these studies revealed that at laser irradiances of

2x10 W/cm2 and greater, a reproducible ablation composition was produced. This was











Table 4-2. Oxide melting points.

Element Oxide Melting Point (C)*

Bi 180,825

Ag 230

Pb 290-500, 886

Cs 400

Rb 400,570

B 450

W 800-900, 1473, 1500-1600

Co 895, 1795

Cu 1235,1326

Mn 1564

Zn 1975

Ni 1984

La 2307

Y 2410

Sr 2430

Ca 2614

Zr 2700

a from reference 79

















(a) F 2.0-(b)
2.0 -- --Rb Fo-
...- A..CS .--Y


LY~bI 1.0-
15 0




S0.5
a 10

0.5.
4 -, 0.0 -- i- i
I 1 E8

Laser Irradiance (W/cm2) Laser Irradiance (W/cm2)





Figure 4-2. Effect of laser irradiance on relative analyte signals for
elements with (a) lower oxide melting points and (b) higher
oxide melting points than Mn.








important considering that slight changes in the laser focus, resulting from crater

formation, surface roughness, or the analysis of samples with small differences in height

would not significantly affect LA-ICP-MS results. The case is different when spatial

information is sought, since the crater diameter and depth are dictated by the laser

irradiance. In this instance, ablative fractionation would be more significant and proper

consideration would have to be given to account for this effect ifanalyte concentrations

were sought.

The consistency of the ablation product did not guarantee that it was truly

representative of the bulk material. Some melting of the glass material did occur as

evidenced in Figure 4-3, which compares the unablated glass surface (a) to the inside of a

laser produced crater (b). This and all other scanning electron micrographs (SEM's) were

taken with a JEOL 35CF electron microscope. The representativeness of the ablation

product would be assessed from solution calibration results.

Solution-Based Calibration

In the discussion on the solution calibration procedure, it was stated that accurate

results could only be obtained if two criteria were met: (1) identical plasma conditions

should be maintained for both solution and ablated solid analysis and (2) the ablation

process should produce a representative subsample. The latter aspect was recently

discussed. The criterion of identical plasma conditions was speculated to exist based on

similar RF optimization behavior for ablation and solution measurements (see Figure 3-

2). This point was more clearly illustrated when RSF's obtained from a 10 ppb

multielement solution and laser ablated NIST 611 glass with the ICP-MS operated in







(a)


(b)


Figure 4-3. Comparison of (a) unablated and (b) ablated glass surface.





















Table 4-3. Comparison of RSF's for solution and laser ablation measurements.

Element Solution RSF8 LA-Wet RSFI LA-Dry RSFa
(sd)b (sd)b (sd)b
Mn 1.00 (.03) 0.87 (.04) 0.41 (.02)
Ni 0.20 (.01) 0.19 (.02) 3.1 (.5)
Cu 0.51 (.01) 0.44 (.04) 0.46 (.04)
Zn 0.064 (.005) 0.046 (.003) 0.14 (.02)
Sr 0.607 (.02) 0.45 (.03) 1.6 (.2)
a relative to Co
b standard deviation for n=5 measurements








both the dry and wet plasma (dual sample introduction) modes were compared (Table 4-

3). All results were relative to Co, which was present in the glass at around 390 ppm.

The RSF's obtained from solution and those for laser ablation with dual sample

introduction mode agreed very well with one another (5-28 %), especially considering

that a noncertified minor element was being used as the internal standard. This indicated

that under the experimental conditions used, the glass matrix was representatively

sampled. Laser ablation measurements with a dry plasma resulted in considerably

different RSF's compared to solutions (10-1450 %), even though identical laser sampling

conditions were employed. The discrimination, therefore, arose in the ICP-MS and

indicated the importance of matched plasma conditions for obtaining accurate results.

The need to measure an internal standard was somewhat restrictive; however, it

was extremely beneficial in providing stable and reproducible measurements. This was

not only in terms of correcting for differences in mass ablated, but also in accounting for

instrumental drift. This point is clearly illustrated in Figure 4-4. In this plot, the average

intensities of 59Co and 60Ni from a 50 ppb solution were plotted over a series of fifteen

scans (1 min/scan). Even though the solution was being continuously introduced at the

same rate, a significant decrease in signal levels was observed. It was later determined

that this was due to an aging detector. The use of absolute intensities would have been

meaningless in this case, but the relative intensities remained stable (1.2 % rsd) since all

isotopes exhibited the same behavior.

Similarly, it was observed that RSF's obtained from solutions did not change

significantly over time due to similar ICP-MS tuning conditions. This was somewhat














1.2x10


.LOx10


- 1.0-
S0.9"
o o.s.
- 0.7.
S0.6.
S0.5.
0.4.


Ni/Co rsd = 1.2 %


I0.31
2-- 8.0x10 0.2 4 8 1 14 1
.Ni Scan Number
rNi
o U


6.0x105 0 U U




4.O xlO 40
** 0.,,

4.Ox10 .
.



2.0x1I0 5_ __________
I I I I I I II I I
0 2 4 6 8 10 12 14 16


Scan Number


Figure 4-4. Ni and Co intensities as a function of scan number (time).
Inset shows the ratio of the two intensities.








surprising, even though the same tuning procedure was used each time the instrument

was run. The implication was that RSF's measured on one day were reliable for

considerable periods of time if no major instrumental modifications were made. To

illustrate this point, the element concentration values from a NIST 611 glass determined

by the solution calibration procedure were compared for three different days. The

solution RSF's determined on the first day were used to calculate analyte concentrations

(see Equation 4-3) on that day, the next day, and one month later. The results are given

in Table 4-4 and indicate that significant differences in accuracy were not observed over

this period of time. The precision (% rsd, n=5 replicates) of the measurements was

typically 10 % or better. Considering that a trace element (Sr) was used as the internal

standard, the accuracy and precision of these measurements was quite good. In later

measurements, an isotope of a major matrix constituent (43Ca) was used as the internal

standard. For all of these measurements, ablation and data collection were performed at

five different fixed locations on the glass surface. After this time, the instrumentation

was modified to allow for translation of the sample during ablation. A comparison of the

results obtained for these two different sampling modes will be presented.

Spot Sampling Versus Line Sampling

Sample translation during ablation was investigated because of several potential

advantages. Sample translation should have provided higher signals since a partially

fresh surface was being ablated by each laser shot and would be expected to give more

representative sampling for inhomogeneous materials. In addition, reduced fractionation

should have resulted since the laser was not defocused from crater formation and



















Table 4-4. Results for NIST 611 glass based on solution calibration.


Cert. Conc.' Meas. Conc. Meas. Conc. Meas. Cone.
(ppm) (ppm) (ppm) (ppm)
Day 1 Day 2 Day 30
(% difference) (% difference) (% difference)
Mn 485 470 (-3.1%) 454 (-6.3 %) 354 (-27%)
Ni 458.7 489(6.6%) 438 (-4.4%) 339 (-26%)
Co (390) 389 (-0.2%) 447(14%) 312 (-20%)
Cu (444) 366 (-17%) 350 (-21%) 342 (-23 %)
Zn (433) 312 (-28%) 295 (-32%) 269 (-37%)
Rb 425.7 399 (-6.3%) 413 (-3.0%) 434(1.9%)
T1 (61.8) 50.0(-19%) 66.5(7.6%) 72.1(17%)
Pb 426 366(-14%) 415(-2.6%) 464(9.0%)
U 461.5 352(-24%) 445(-3.7%) 497(7.8%)
a Values in parentheses are not certified








localized sample heating would be less significant since the same area was not

continuously ablated with the high powered laser.

For a typical 1 minute analysis (300 laser shots), translation of the sample at 15

p.m/s produced signals that were on average 2 to 3 times higher than those obtained from

a single spot. The dimensions of a typical single spot and translated sample crater were

measured with an optical microscope. For single spot sampling, the crater was estimated

to be 75 pm wide and 120 gpm deep. If a parabolic crater was assumed, this corresponded

to a total ablated mass of 830 ng (2.2 ng/shot). For translation, the crater was 50 gm

wide, 15 p.m deep, and 900 p.m long (assumed from 15 pm/s x 60 s). If a parabolic

trough was assumed, this corresponded to a total ablated mass of 1350 ng (3.8 ng/shot).

This provided further evidence that translation resulted in higher ablation efficiencies.

The exact magnitude of signal enhancement could not be inferred from these

measurements, since the sample was typically ablated for 100 shots (20 s) before signals

were acquired. It was previously shown (Figure 3-8) that the ablation efficiency

decreased significantly after the first 50 shots when single spot sampling was used.

NIST 611 glass was analyzed to assess the homogeneity of the material and to

determine if sampling strategy had any effect on the accuracy of solution calibration

based measurements. For this study, scans were made at different spots (25 total) on the

sample or at different lines (5 scans/line) produced from translating the sample. Several

isotopes (43Ca, "55Mn, 59Co, 60Ni, 63Cu, 88Sr, 208Pb) were measured for a total acquisition

time of one minute. Ca served as the internal standard in these measurements, since it

was a major matrix constituent (12 % CaO). The RSF's determined from both sampling

methods were compared with RSF's obtained from a solution containing 20 ppb of all the








elements of interest, except for Ca, which was present at 2 ppm. This closely matched

the ratios of trace element to Ca in the glass sample. The results for Mn, Co, Cu, and Sr

are presented in Figure 4-5. The translation values were significantly higher for Mn and

Co, slightly higher (statistically significant at 95 % level) for Cu and Ni (not shown), and

almost identical for Sr and Pb (not shown). Where differences did occur, spot sampling

measurements were more accurate as evidenced by their proximity to the solution results.

The most likely reason for these discrepancies was either inhomogeneity in the sample or

the presence of fractionation.

Examination of the glass surface with scanning electron microscopy revealed the

presence of a significant amount of redeposited particles (Figure 4-6). For comparison,

the top of this picture contains areas that had been previously ablated. After ablation, the

sample surface was wiped clean to remove any redeposited material. It was thought that

these redeposited particles might possibly be enriched or depleted in certain elements and

could influence the laser based RSF's, making them vary more significantly compared to

the solution value. The effect would be more significant for translation analyses since

areas where particles were redeposited would be continuously sampled as the laser

probed the surface. Also, because the surface layer sampled was smaller with translation

the proportion of redeposited material to fresh material would be greater in this case. In

order to investigate these potential sources of variation, a larger set of elements was

studied and sampling was performed in a way that would maximize any effects due to

sampling of redeposited material.

In this study, elements encompassing a wide range of oxide melting points (see

Table 4-2) were measured since it has been shown that the extent of fractionation












800-r (a)


750


700


650


600


550


I I I *
-I


Solution(avg) = 574, sd = 9 (1.5% rsd)
Spot(avg) = 584, sd = 26 (4.5% rsd)
Line(avg) = 668, sd = 25 (3.8% rsd)


0 5


10 15 20 25
Scan Number


\ Line(avg) = 477, sd = 7 (1.4% rsd)
(b Spot(avg) = 557, sd = 26 (4.8%rsd)
Line(avg) = 645, sd = 28 (4.4% rsd)





L A -ln e k "



L spo't,

.S!lufion


5


10 15 20 25
Scan Number


Figure 4-5. Comparison of results for spot and line sampling for Mn, Co, Cu, and Sr.




















Solution(avg) = 233.4, sd = 5 (1.9% rsd)
Spot(avg) = 254, sd = 15 (5.9% rsd)
Line(avg) = 269, sd = 13 (4.9% rsd)


0 5 10


15 20 25 I I
15 20 25


Scan Number



Solution(avg) = 230, sd = 6 (2.6% rsd)
Spot(avg) = 234, sd = 10 (4.3% rsd)
Line(avg) = 232, sd = 11 (4.6% rsd)


5U M I I


10 15

Scan Number


20 25


Figure 4-5 continued.


(c)









LA-line
I


5


zUU I"t1111"" ._ ____





























.. ." ":*" -
* P '
.7'. : "


Figure 4-6. SEM of glass surface after ablation.








correlated with this property (75, 78). The isotopes used were 43Ca, "Mn, 59Co, 6Ni,

63Cu, 66Zn, 8Sr, 9Zr, 7Ag, 139La, and 208Pb. The oxide melting points ranged from 230

C for Ag to 2700 C for Zr. Both single spot and translation measurements were

performed as previously described. In order to maximize the proportion of redeposited

material sampled, additional analyses were performed close to previous ablation tracks at

the same laser energy (0.5 mJ) and with lower energies (0.1 mJ). The use of lower laser

energies should result in the sampling of an even smaller surface layer. To ensure that

any differences observed at low energies were due to the sampling of redeposited

particles, analyses were also performed at a location well removed from any previous

ablation craters.

Table 4-5 summarizes the results obtained for glass ablated with single spot

sampling, translation sampling, and translation sampling through areas with significant

amounts of redeposited material. All of these measurements were performed with a laser

energy of 0.5 mJ. Significant differences (> 10 %) between spot and translation sampling

were observed for Mn, Co, Zr, and La. A statistically significant difference (at 95 %

level) was also observed for Ag. RSF's for Mn and Co increased with line sampling,

while those for Zr, La, and Ag decreased. No major difference between translation

measurements was observed in this study, indicating that the particles did not have any

effect on the results at this laser energy. The differences observed for spot and translation

sampling could not be explained on the basis of fractionation. If fractionation was

occurring, it should have been more prevalent with spot sampling because localized

heating would be more likely in this case. Since Ag and Zr had the lowest and highest

oxide melting points, respectively, they should have displayed opposite behaviors. The

















Table 4-5. RSF's obtained from various sampling strategies.


Single Spot Translation Translation-redep.
RSF' (sd)b RSFA (sd)b RSFa (sd)b

Mn 530 (20) 610(30) 620 (30)
Co 510(20) 580(30) 580(30)
Ni 122(5) 121(5) 123(5)
Cu 230(9) 240(10) 250(10)
Zn 56(3) 53(3) 54(3)
Sr 217(5) 218(8) 216(7)
Zr 139(5) 113(5) 113(2)
Ag 120(5) 112(4) 115(5)
La 101(6) 89(3) 84(4)
Pb 1.6(.1) 1.8(.2) 1.8(.2)


a relative to 43Ca
b standard deviation (n =


15 measurements)








RSF for Ag should have increased for spot sampling (observed), while Zr should have

remained stationary or decreased slightly (opposite observed). In addition, Mn and Co

exhibited the largest RSF changes, even though these elements did not possess melting

points significantly different from elements (e. g. Cu, Zn, and Ni) where no changes in

RSF's were observed. The differences observed were most likely the result of small-

scale inhomogeneity in the glass samples. Measurements performed with a lower laser

energy supported this notion.

A comparison of RSF's (Figure 4-7) obtained for spot sampling with a laser

energy of 0.5 mJ and translation sampling with energies of 0.5 mJ and 0.1 mJ revealed

significant differences. In this plot, RSF values were normalized to solution RSF values;

therefore, the accuracy of the measurements was readily apparent. For example, a value

of 1.1 indicated a 10 % error, 1.2 a 20 % error, and so on. The low energy measurements

were performed both near a previous ablation track and well beyond any previously

ablated areas. No significant difference between these measurements was observed,

providing more proof that the particles did not affect the accuracy of measurements. For

Ag and Cu, these low energy measurements resulted in significantly higher RSF's

compared to higher energy measurements. This was almost certainly due to

inhomogeneity since an even smaller surface layer was being sampled in these cases. It

could have resulted from surface contamination, but this was unlikely since only these

two elements demonstrated this behavior. Likewise, if fractionation were at fault

differences would have been observed for other elements as well.

The results clearly indicated the importance of introducing a representative

portion of the sample during analysis. Single spot sampling resulted in better accuracy

















0
S1.4
0


0-
I


Z 1.0


~0.8


0.6


0.4


02
Mn Co NI Cu Zn Sr Ag La Pb




Figure 4-7. Comparison of RSF's obtained for spot sampling (0.5 mJ) and line sampling (0.5 mJ and 0.1 mJ).








(average of 9.5 %) than translation measurements (average of 14 %) for glass, since it was

less affected by in-depth inhomogeneity. Sampling depths were around ten times higher

with the former. The precision (% rsd) of the measurements was less than 5 % for both

sampling modes. Lateral inhomogeneity was not a problem in either case, due to the

relatively large crater diameters (50 mrn or greater). This behavior should not be

generalized since the sampled volumes, homogeneity, and physical characteristics of the

sample all play a role in determining the most suitable sampling strategy.

Conclusions

Calibration using standard solutions and dual sample introduction was shown to

provide reasonably good accuracy (+/- 10 %) for trace elements in NIST glass samples.

Detection limits (3a) were less than 1 ppm for all elements studied. The effects of

ablative fractionation were studied and it was determined that at irradiances > 2x 108

W/cm2, representative sampling was achieved. Sampling strategies were compared and it

was determined that spot sampling produced more accurate measurements for the glass

samples. Translating the sample produced higher signal intensities; however, local

heterogeneity was more problematic because a smaller surface layer was sampled in this

mode. Ablated mass was estimated to be a couple of ng's per shot.














CHAPTER 5
ANALYSIS OF SILICON NITRIDE CERAMIC BEARINGS

Introduction

There is considerable interest in using silicon nitride (Si3N4) bearings for a wide

variety of applications. Such interest results from the unique chemical and physical

properties which these ceramics possess. In comparison to conventional steel bearings,

silicon nitride bearings offer high speed and acceleration capability because of their low

density, extended temperature capability, longer lifetimes and lower wear rates, excellent

corrosion resistance, and the ability to operate under conditions of marginal lubrication

(80). This combination of qualities has led to investigations of silicon nitride bearings for

use in high speed, high temperature applications, such as in the aerospace industry.

The physical and mechanical properties of ceramic materials are influenced by

trace element impurities (81-84). Therefore, comprehensive trace element

characterization, in terms of both the bulk composition and spatial distribution of

elements, is required for ceramic materials if they are to be used in demanding

environments. Digestion procedures have been used for ceramics with analysis of the

resulting solution by several techniques, including ICP-AES/MS (85, 86) and Flame AAS

(85). These procedures are difficult and time consuming, due to the resistance of ceramic

materials to chemical attack. Wet chemical procedures can also introduce contaminants,

and they only provide information on the bulk sample and tell nothing of the distribution

of elemental constituents in the ceramic.








Direct sampling methods are preferred for the analysis of compact ceramics.

Several reviews discuss the methods currently used for analyzing ceramic materials (81,

82). Characterization of ceramics on both the bulk and micro scale is required in many

cases. Most often, several complementary techniques are necessary to provide this

information. LA-ICP-MS is extremely well suited for the bulk characterization of

ceramic materials, and also possesses the capability of providing spatial information on

the tm scale. The suitability of laser ablation results from the ability to sample

essentially any type of material with little to no sample preparation. The major obstacle

to the use of LA-ICP-MS for the quantitative analysis of ceramic materials has been the

lack of suitable standard reference materials. In some cases, synthetic ceramic standards

were created for analyzing ceramics with LA-ICP-MS (87, 88); however, these were

designed to analyze a specific material. It would be valuable to develop methodology

that would allow for the accurate analysis of a wide range of ceramics. This issue was

the driving force behind the investigation of solutions for calibration of solid materials, as

discussed in Chapter 4.

In this chapter, use of the solution calibration methodology for reliable

quantitative elemental analysis of silicon nitride ceramic bearings will be presented. In

addition, the use of NIST glasses for calibration of silicon nitride ceramics will be

discussed. The results obtained from the two methods were compared to one another,

and with results obtained by electron probe microanalysis (EPMA). Ablation craters

have been characterized using both profilometry and scanning electron microscopy. The

detection efficiency of the LA-ICP-MS system was estimated based on the profilometry

results.








Experimental

The LA-ICP-MS system (Chapter 3) and typical operating conditions (Chapter 4)

were previously discussed. The laser was operated at 5 Hz, with pulse energies of 0.7

mJ. Both single spot and translation sampling were used in this work. Typical analysis

times were 60 to 80 s (signals averaged over 300 to 400 laser shots).

Silicon nitride bearings (NBD-100 Cerbec, East Granby, CT) were mounted in

epoxy and cut with a diamond blade wafering saw to obtain a flat surface. They were

then polished to a 1 p.m finish. These sample preparation steps were required for the

EPMA analysis only. Laser ablation analyses could be performed directly on the intact

bearing.

The solution calibration methodology was presented in Chapter 4. For the

analysis of silicon nitride, it would have been preferred to use a minor isotope of silicon

(29Si or 30Si) as the internal standard directly; however, large interference at these

masses resulted in saturation of the multiplier detector. Magnesium was chosen as an

alternative, since it was present at significant levels (1000s ofppm) in the samples, and

also possessed an isotopic pattern which allowed for its measurement using both the

Faraday (24Mg 79 % abundance) and multiplier (25Mg 10 % abundance) detectors.

The analytical procedure involved measuring 28Si and 24Mg with the Faraday detector and

determining the concentration of magnesium in the sample based on the solution

RSF(24Mg/28Si). The concentration of silicon (60 %) was estimated from sample

stoichiometry. All other elements (minor and trace) were measured with the multiplier

detector and the concentrations were determined from sensitivity factors relative to 25Mg.








In the glass-based calibration work, a National Institute of Standards and

Technology glass was used (NIST 611). Cobalt was used as the internal standard for

these studies, since its concentration in the glass was known and magnesium's was not.

The calibration procedure was similar; RSF's (analyte/59Co) obtained from the glass

matrix were used to calculate the concentrations of analytes in the silicon nitride bearing.

Ablation craters were characterized using an Alpha-Step 500 profilometer

(Tencor Instruments, Santa Clara, CA) and a JEOL 35CF scanning electron microscope

(SEM). For the SEM's, the silicon nitride samples were attached to mounts with carbon

paint and coated with AuPd to provide a conductive surface. EPMA of the silicon nitride

bearings was used for comparison with LA-ICP-MS results. For these analyses, a carbon

coating was used since this resulted in lower background levels. The work was

performed on a JEOL Superprobe 733.

Results

Identification and Distribution of Elements in Silicon Nitride

A complete mass scan of the silicon nitride bearings (NBD 100) used in this study

resulted in the identification of 26 elements (Mg, Al, Ca, Sc, Ti, Cr, Ni, Co, Cu, Sr, Y, Zr,

Mo, Nb, Ag, Cd, Sn, Ba, La, Ce, Pr, Nd, Ta, W, Pb, and Bi). Although no attempt was

made to quantify all of these elements, their estimated concentrations ranged from the

hundreds ofppb level to thousands ofppm. For comparison, a newer grade of bearing

material (NBD 200) was analyzed and contained much lower levels of impurities, with

only the sintering aid (Mg) having a high concentration (> 100 ppm). Only ten elements

were present at high enough levels to be detected with the LA-ICP-MS system. For this








reason, the NBD 100 bearings were chosen for all subsequent work since they provided a

more analyte-rich sample.

For LA-ICP-MS to be used for bulk analysis of the ceramic bearings, it was

essential that the small surface layer sampled (~ 5 4im) on the exterior of the bearing be

representative of the bulk material (Figure 5-1). This was tested by analyzing the interior

and exterior surfaces of a cut bearing. The measured concentrations of all analytes were

almost identical within the measurement uncertainty, indicating that the relatively small

volume sampled by laser ablation of the surface provided information on the bulk

specimen. The SEM (Figure 5-1) revealed that the ablation process resulted in both

melting and fracturing of the silicon nitride surface. Melting was evidenced by the

presence of redeposited spherical particles around the ablation craters, as well as by a

comparison of the unablated (Figure 5-2 (a)) and ablated (Figure 5-2 (b)) silicon nitride

surface. A discussion of the particles produced, and their origins, from laser ablation of

ceramic, glass, and soil samples will be presented in Chapter 6. Fracturing of the silicon

nitride surface was evidenced by the small step-like features outside of the main ablation

track.

As mentioned in the experimental section, an isotope of magnesium was chosen

as the internal standard for most of the quantitative measurements. A requirement for the

use of internal standardization for obtaining accurate results is that the analyte and

internal standard be distributed similarly over the area sampled, and that they exhibit

similar behavior in transport to and excitation in the ICP. This was tested, albeit on a

limited basis, by measuring the temporal response of the 25Mg and 63Cu signal over a

period of 90 s, while the laser was repetitively fired (Figure 5-3). The two signals









































-- ----.--
*, -^ .^ ^ ..H .. .. ^ '
..^ y ^ ^ A^-S'. ^ .. ,',' ".':" ''" ; :' ?*':, .7l:^


Figure 5-1. Scanning electron micrograph of a laser ablation track in
silicon nitride. The sample was translated at 15 pnm/s.
SEM was taken with the sample tilted at 50 degrees.
















































Figure 5-2. Scanning electron micrographs of (a) unablated and (b)
ablated silicon nitride surface.

































I5m


63CU


I I I I I I I I I


10 20 30 40 50

Time (s)


60 70 80 90


Figure 5-3. Temporal profiles of Mg and Cu signals from silicon nitride.








correlated very well with one another, and partially justified the use of 2Mg as the

internal standard for quantitative measurements of trace constituents in the ceramic.

Additionally, the distribution of magnesium and titanium was studied in the silicon

nitride bearing. Signals for these two elements were measured from 24 different

locations on the sample. The % rsd of the elemental ratios (Ti/Mg) was 10 %, indicating

that titanium and magnesium were distributed similarly over the surface of the bearing.

Results from Solution- and Glass-Based Calibration

Results for the analysis of a silicon nitride bearing using RSF's obtained from

standard solutions are given in Table 5-1. For comparison, results obtained with EPMA

are provided for the three most concentrated elements (Mg, Co, Al) present in the silicon

nitride. Because of the inferior sensitivity of EPMA, these were the only elements

detectable in the sample by this method. The agreement between the techniques was

excellent, suggesting that the use of solutions for calibration is effective for the analysis

of ceramic materials. The precision of the techniques was similar, with % rsd's (n=10 for

LA and n=5 EPMA) ranging from 2 to 18 %.

In addition to solution-based calibration for silicon nitride ceramics, the use of

NIST 611 glass as a calibration standard was investigated. In this work, a second bearing

was analyzed. Similar to the procedure used for solutions, RSF's (analyte/Co) obtained

from the glass sample were used to determine the concentration of analytes in the ceramic

sample. A smaller number of elements was used in this study because of the limited

number of certified elements in the glass. In several cases, the noncertified values

provided by NIST were used (Co, Cu, and Zn). The results are presented in Table 5-2















Table 5-1. Comparison of results obtained with LA-ICP-MS using
solution calibration and EPMA.

Analyte LA-ICP-MSa EPMAb
Concentration (ppm) Concentration (ppm)

Mg 7200 +/- 400 7300 +/- 500

Co 1420 +/-60 1300 +/-200

Al 2230 +/-60 2400 +/-50

Ni 79+/-6

Cu 28+/-3

Sr 8+/-1

Mo 12+/-1

Ba 16+/-1

La 7+/-1

Nb 38+/-6

Y 18+/-2

a Confidence interval (95 %) based on 10 measurements.
b Confidence interval (95 %) based on 5 measurements.


















Table 5-2. Comparison of results using solution and glass-based calibration.

Analyte Results from Results from Results from
solution RSF's glass RSF's glass sensitivities
(ppm)8' b (ppm)a, b (ppm)8
Mn 103+/-3 123+/-5 120+/-15

Co 1540+/-60 1380+/-90 1400+/-200

Ni 58+/-2 65+/-3 63+/-8

Cu 9+/-2 10+/-2 9+/-2

Zn 9.2+/-0.7 13+/-1 13+/-2

Sr 7+/-1 9+/- 1 8 +/-2

a Confidence intervals (95 %) based on 10 measurements.
b Results for Co based on Co/Ni relative sensitivity factor (RSF).








and demonstrated that similar results were obtained for both solution- and glass-based

calibration.

The last row of this table represents an extension of the use of glasses for

calibration. In this case, the absolute analyte sensitivites (intensity / concentration) from

laser ablated glass were used with a correction factor to account for differences in the

ablated mass between the glass and ceramic sample. The value of the correction factor

was determined by measuring the 28Si from both samples and taking into account the

differences in nominal silicon concentration of the samples. It was determined that the

ablated mass was four times greater for the glass than the ceramic, under similar

sampling conditions. Not surprisingly, the measured analyte concentrations agreed with

those obtained by the other methods. The precision of these measurements was worse

(12-15 % rsd), since the measurements involved non-normalized sensitivity values

obtained from glass.

It would have been beneficial to measure an isotope of silicon along with the

analytes to directly compensate for variations in ablated mass; however, this was not

feasible with the present system. The use of solutions produced high background levels

(mainly oxides) and precluded the measurement of 29Si and 30Si with the multiplier

detector. This could be remedied, at least in part, by incorporating a desolvation

apparatus to further decrease the solvent loading; however, this was not attempted in this

study. Alternatively, independent measurements of ablated mass, such as the use of

acoustic waves generated by the ablation process (89), or measurement of the scattering

signal produced by the mobilized ablation product (90) could prove useful for the

analysis of ceramic materials with glass-based calibration. The use of independent mass








normalization would eliminate the need for internal standardization and provide a more

powerful method for analyzing ceramic materials based on glass standards. Light

scattering for mass normalization with glass, ceramic, and soil samples will be presented

in Chapter 6.

Measurement of Mass Ablated and Estimation of System Efficiency

Profilometry and weight loss measurements were performed on the silicon nitride

samples to determine the amount of material removed by the ablation process and to

determine the efficiency of the LA-ICP-MS system. As previously addressed (Chapter

3), sample translation resulted in a larger mass removal rate and subsequently higher

sensitivities. Profilometry measurements (Figure 5-4) on the craters shown in Figure 3-8

indicated that the volume ablation rate was nearly double in the case of sample translation

compared to single spot sampling (3 pL/s compared to 1.6 pL/s). These corresponded to

mass ablation rates of 9 ng/s and 5 ng/s, respectively. The higher ablation rate was due to

the fact that a partially pristine surface was being ablated by each laser shot with

translation of the sample. The ablated masses agreed reasonably well with those obtained

by weighing the sample before and after approximately 50,000 laser shots. The observed

ablation rates were used to determine a system efficiency for the sintering aid Mg (Mg

ions detected / Mg atoms removed). In these measurements, the 24Mg signal was

continuously monitored during the ablation process. The total integrated signal was then

divided by the estimated mass of Mg ablated. Efficiency of detection for Mg was

estimated to be 1 in 500,000. The efficiency of detection for analytes is a function of the

ablation process, transport process, ionization in the ICP, and mass spectrometer

response.









Translation at 15 um/s


E
r. -4
4 -4-
c -6
0J)
E -8
-10 Tencor Alpha Step 500 Profilometer
CS -10 ~____
) 2 50 100 150
Q
- 0
o *50 shots without translation
: -2
(U)
-4
-6
-8
-10
I I I I
50 100 150

Horizontal Displacement (prm)


Figure 5-4. Profilometry traces of craters in silicon nitride with and without translation of the sample.








Conclusions

The use of both solution and glass-based calibration has been demonstrated for

the analysis of silicon nitride bearings. The results obtained from solutions were verified

with a complementary technique, EPMA. Use of glasses for calibration provided similar

results to those obtained from solutions and may prove useful in the future for a wide

variety of ceramic materials. A large number of standard glass materials are presently

available, making them very attractive as potential calibration materials for ceramics.

The ability to independently measure differences in the mass ablated for glass and

ceramics would greatly improve the utility of glass-based calibration.














CHAPTER 6
INVESTIGATION OF LIGHT SCATTERING
FOR MASS NORMALIZATION IN LA-ICP-MS

Introduction

To account for variations in ablated mass and improve the precision and accuracy

of laser ablation measurements, the signal from a matrix element is commonly used as an

internal standard (38, 45, 46, 70). This improves the precision, provided the spatial

distribution of the internal standard and the analyte are similar, and the elements behave

similarly in terms of ablation, transport, and ionization in the ICP. To improve the

accuracy of LA-ICP-MS measurements, it is also required that the concentration of the

internal standard be known. In many cases, this information is not readily available or

can vary significantly, such as with many geological materials and multi-layered

ceramics. In such instances, there is a need for independent measurement of the mass

ablated.

Several techniques that independently measure differences in the mass ablated

have been investigated (89-91). Pang et al. demonstrated that the acoustic wave

generated by the ablation process could be used to normalize analyte signals in LA-ICP-

MS (89). Their work resulted in modest improvements of the precision compared to the

non-normalized signals from steel and aluminum alloy samples. Other work has

involved measurements on the mobilized ablation products. Richner et al. measured the

light loss caused by scattering from ablated material as it passed through a cylindrical








glass tube (91). They reported that normalization ofanalyte signals from cast iron

samples based on this technique was comparable to the use of an internal standard.

Similarly, Tanaka et al. used the light scattering signal produced from the transported

aerosol to normalize analyte signals (90). In their work with zirconium alloys, they also

reported an improvement in precision that was comparable to internal standardization.

In this work, the use of light scattering for normalization with several matrices,

including brass, glass, soil, and Macor (Coming Glass Works, Corning, NY) ceramic

has been investigated. It is demonstrated that a simple light scattering system can be used

for mass correction with a variety of materials. Normalization based on light scattering

results in improved precision of LA-ICP-MS measurements; however, this method of

normalization is not as good as the use of an internal standard. The major strength of the

present technique is that it did not require any knowledge of the sample homogeneity or

concentrations of elements. This makes it potentially useful for the accurate analysis of

inhomogeneous samples as well as those for which standards are not readily available.

This latter aspect has been studied by examining the effectiveness of the light scattering

system for normalization between different types of samples (glass, ceramic, and soil).

Experimental

The LA-ICP-MS instrumentation (Chapter 3) and operating conditions (Chapter

4) were previously described. One difference was the use of a smaller ablation cell. It

consisted of a 3 cm i.d. Plexiglas tube with a quartz window for transmission of the UV

laser beam. The total volume of the new ablation cell was 50 cm3. The smaller cell

was used to produce a denser ablation aerosol for larger scattering signals.








The scattering cell (Figure 6-1) was placed in-line with the transport tubing. It

was placed approximately 0.5 m from the ablation chamber. The body of the scattering

cell was a 4-way Swagelok(Crawford Fitting Co., Solon, OH) connector with a 1.5 mm

hole drilled in the top for measurement of the particle scattering at 90. Smaller angle

viewing would result in higher scattering intensities, but the present arrangement was

chosen for its simplicity. Windows of the scattering cell were mounted at Brewster's

angle on 3 cm extension arms made of stainless steel tubing. The arrangement produced

less stray-light scatter from the windows and interior of the cell. A 3 mW polarized

HeNe laser (Aerotech, Inc., Pittsburgh, PA) operating at 632.8 nm was used as the

scattering source in these measurements. To improve the quality of the output beam, the

laser was spatially filtered before the scatter cell with an adjustable iris.

An end-on photomultiplier tube (R647, Hamamatsu Photonics K.K., Japan) was

used to collect the scatter produced as the ablation aerosol passed through the cell. The

entire scatter assembly was enclosed in a light-tight box. An alternative arrangement,

which would eliminate the need for darkness, would be to use an interference filter that

would transmit only a very narrow region of light centered at the HeNe wavelength.

Signals from the PMT were amplified by a current to voltage amplifier (Keithley

Instruments, Inc., Cleveland, OH) with a gain of 106 V/A and a rise time of 100 ms. The

amplifier output was fed to a computer interface module (Stanford Research Systems,

Inc., Palo Alto, CA) which was triggered at 100 Hz. Normalization measurements were

made by integrating the scatter signal over a given number of shots, as well as averaging

the "steady-state" level produced during continuous ablation of translated samples for a

total of 200 shots (40 s).









(a)



From Ablation
Chamber











(b)

From Ablation
Chamber


Beam Dump


PMT


To ICP-MS


---- Iris


He-Ne Laser


PMT


To ICP-MS


Window


Figure 6-1. Schematic of light scattering set-up ((a) top view and (b) side view).








NIST glass (611, 612, 614, and 617) and brass (c110l, 1102, c1109, cl 10)

standard reference materials were used to assess the utility of the scattering system for

normalization ofanalyte signals within a particular matrix. In the glass samples, the 8Sr

isotope was measured with the ICP-MS as well a minor isotope of a major matrix

constituent (43Ca) in order to compare the results from scatter normalization with those

obtained by internal standardization. The concentration of Sr in the glasses ranged from

41.72 ppm to 515.5 ppm. For the brass samples, the 68Zn isotope was measured.

Concentrations of Zn in these standards ranged from 15.2 % to 30.3 %. In the inter-

matrix studies, NIST 611 glass, NIST 2704 soil, and a Macor ceramic disc were used.

The soil samples were pressed into pellets, without the addition of a binder at a pressure

of 35 MPa, for laser ablation analyses. In these studies, the 28Si isotope was measured

since silicon is the major matrix element in each of the matrices. The amount of silicon

in the glass, Macor, and soil was 34 % (nominal), 21.5 % (nominal), and 29.66 %

(certified), respectively.

The nature of the particulate material produced from laser ablation of glass, soil,

and Macor samples was studied by collecting ablated material on 0.3 tun membrane

filters (Millipore Corp., Bedford, MA). Because of the significant pressure drop created

with the in-line filters, it was necessary to use a vacuum pump on the back side of the

filter to eliminate flow restrictions. Flow conditions, identical to those used in the

normalization experiments, were achieved by adjusting the pump valve and measuring

the Ar flow rate with a rotameter. After collection, filters were glued onto mounts with

conductive carbon paint and coated with AuPd for examination with a scanning electron

microscope.








Results

Scatter Signal Normalization and Comparison with Internal Standardization

Initial experiments with the scattering system were performed to determine if the

scattering signals correlated with signals from the ICP-MS. Figure 6-2 illustrates typical

(a) scatter and (b) mass spectrometric signals obtained from 50 laser shots on NIST 611

glass. It should be mentioned that the actual transit time to the scattering cell was -1 s,

and -3 s to the ICP-MS. The large number of spikes present in the scatter signal was

assumed to be due to larger particles passing through the scatter cell based on visual

observation. To evaluate the efficiency of the system for detecting changes in the mass

removed, ablation was performed with different laser energies and focus positions. The

total difference in mass ablated over these conditions ranged over approximately one

order of magnitude. A plot of the average scatter and 28Si mass signals (Figure 6-3)

demonstrated that the two signals were highly correlated (R = 0.98). In this example,

differences in the mass ablated were generated by changing the laser sampling parameters

(i. e. focus and pulse energy); however, these differences occurred naturally over the

series of glass standards under identical sampling conditions. This was probably due to

differences in the surface characteristics of the glasses, since they were all completely

opaque at 266 nm. Similar behavior has been reported by Mermet and co-worker in the

laser ablation of glasses with a 266 nm Nd:YAG (92).

In the absence of normalization, differences in the mass ablated between the

samples would not allow small changes in analyte concentration to be detected. This is

illustrated by a series of calibration curves for Sr in NIST glasses (Figure 6-4), where in

(a), the non-normalized analyte signals are plotted; in (b), the scatter normalized signals







91
















0.8
(a)
0.6

0.4

0.2

4) 0.0

-0.2
0 2 4 6 8 10 12 14 16 18 20 22

2.0x10 7 Time (s)
(b)
S1.5X1O 7

.J l.OxlO 7


5.OxlO 6

0 .0 --- .

0 2 4 6 8 10 12 14 16 18 20 22

Time (s)






Figure 6-2. Plots of(a) scatter and (b) mass spectrometric signals
obtained from 50 laser shots on glass surface.













7x10 8



6x10 8 *



5x10 8

o U
W 4x10 8



3xlO 8-



20 8
2x10 8


1x10 8_s/



0_

I I I I I
0 50 100 150 200
Scatter Signal (a.u.)



Figure 6-3. Correlation plot of mass spectrometric and scatter signals from NIST glass.




Full Text
ANALYSIS OF GLASS, CERAMIC, AND SOIL SAMPLES USING
LASER ABLATION INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY
By
SCOTT A. BAKER
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
1998

ACKNOWLEDGMENTS
I would like to thank my research advisor, Dr. James D. Winefordner, for his
patience, guidance, encouragement, and enthusiasm. I consider myself extremely
fortunate to have had the opportunity to work under his direction. I would also like to
thank Dr. Benjamin W. Smith for many useful discussions concerning this research and
for always having an open door. Both of these people were integral in my development
as a scientist.
I am grateful to all of the members of the Winefordner research group for their
friendship. They were a source of inspiration and helped me a great deal in producing
this work. I would especially like to acknowledge those who contributed directly to my
experiments. These people included Igor Gornushkin, David Rusak, Bryan Castle, Robin
Russell, Melody Bi, Rebecca Litteral, Matthew Dellavecchia, and Ricardo Aucelio.
Additionally, I would like to acknowledge Dr. Kobus Visser, Dr. Oleg Matveev, and Dr.
Nico Omenetto for many useful discussions during the course of my graduate studies.
I thank Jeanne Karably for answering many questions and for making sure things
went smoothly. Also, I would like to acknowledge Steve Miles for his electronics
expertise, and Joe Shalosky, Gary Harding, and Dailey Burch of the machine shop for
helping me on many occasions.
I am grateful to Dr. Paul Holloway and Dr. Steve Pearton for use of their
profilometers, and to the staff at the Major Analytical Instrumentation Center for
11

allowing me to use their facilities. They also produced the electron microprobe data.
My parents have always provided me with love and support. They have been a
great source of strength and encouragement throughout my life. I cannot express enough
how much the love, support, and friendship of my wife, Amy, means to me each and
every day. She has always been there to keep me going, and to keep things in
perspective. I look forward to our spending a long and joyous life together.
Finally, I would like to acknowledge the National Science Foundation-
Engineering Research Center for Particle Science and Technology and the Air Force
Office of Scientific Research for funding this research.
in

TABLE OF CONTENTS
ACKNOWLEDGEMENTS ii
ABSTRACT vii
CHAPTERS
1 INTRODUCTION 1
2 BACKGROUND 6
Inductively Coupled Plasmas 6
Inductively Coupled Plasma-Mass Spectrometer Interface 10
Ion Optics 12
Mass Analyzer 13
Detection of Ions 14
Sample Introduction in ICP-MS 14
Laser Ablation 16
3 GENERAL CONSIDERATIONS AND OPTIMIZATION
OF EXPERIMENTAL PARAMETERS 19
Introduction 19
Instrumentation 19
Samples 21
Results 22
Plasma Operating Conditions 22
Scan Parameters and Internal Standardization 33
Sampling Strategy 35
4 ANALYSIS OF GLASS SAMPLES 41
Introduction 41
Experimental 42
Calibration Procedure 43
Results 47
Fractionation 47
Solution-Based Calibration 53
IV

Spot Sampling Versus Line Sampling 58
Conclusions 69
5 ANALYSIS OF SILICON NITRIDE CERAMIC BEARINGS 70
Introduction 70
Experimental 72
Results 73
Identification and Distribution of Elements in Silicon Nitride 73
Results from Solution- and Glass-Based Calibration 78
Measurement of Mass Ablated and
Estimation of System Efficiency 82
Conclusions 84
6 INVESTIGATION OF LIGHT SCATTERING
FOR MASS NORMALIZATION IN LA-ICP-MS 85
Introduction 85
Experimental 86
Results 90
Scatter Signal Normalization and Comparison
with Internal Standardization 90
Comparison of Scatter Normalization for Glass,
Soil, and Macor® Ceramic 96
Conclusions 103
7 ANALYSIS OF SOIL AND SEDIMENT SAMPLES 104
Introduction 104
Experimental 105
Results 108
Spot Sampling Versus Line Sampling 108
Solution-Based Calibration 110
Speciation Effects 117
Effect of Organic Content 121
Analysis of Particle Size Fractions 125
Analysis of Soils Using Standard Additions 130
Detection Limits 134
Conclusions 134
8 EVALUATION OF A COMPACT LASER SOURCE 136
Introduction 136
Experimental 137
Results 138
Conclusions 146
v

9 CONCLUSIONS AND FUTURE WORK 147
Conclusions 147
Future Work 150
REFERENCES 153
BIOGRAPHICAL SKETCH 159
vi

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
ANALYSIS OF GLASS, CERAMIC, AND SOIL SAMPLES
USING LASER ABLATION INDUCTIVELY COUPLED MASS SPECTROMETRY
By
Scott A. Baker
May 1998
Chairman: James D. Winefordner
Major Department: Chemistry
Laser ablation (LA) as a direct solid sampling method for inductively coupled
plasma mass spectrometry (ICP-MS) has been used for the analysis of glass, ceramic, and
soil samples. The strength of this technique is that it can be applied to essentially any
solid material and eliminates the need for difficult and time consuming dissolution
procedures. The major limitation of LA-ICP-MS is the presence of matrix effects,
making quantitation difficult in the absence of standards of identical composition.
This work was largely concerned with the development of methodology for
obtaining accurate quantitative results from a variety of solid materials, including glasses,
ceramics, and soils without the need for matrix-matched standards. The use of solution-
based calibration for the analysis of a National Institute of Standards and Technology
(NIST) glass resulted in excellent agreement (typically +/- 10%) with the certified values.
Detection limits were in the sub-ppm range for all elements studied. The solution
Vll

calibration technique was also applied to silicon nitride ceramic bearings and the
accuracy of the results were confirmed with X-ray microanalysis. In addition, standard
glass materials were useful in the analysis of ceramics provided that differences in
ablated mass were properly accounted for. To this end, the use of light scattering for
measuring the amount of ablated material was evaluated. Light scattering was effective
for mass normalization provided that the particle sizes of the ablation products were
sufficiently similar.
The use of solution-based calibration for soil samples resulted in poorer
agreement than in the case of glasses and ceramics; however, results were still within +/-
20 % for most analytes studied. Studies involving sample particle size, organic content,
and element speciation were performed to understand the effects that these variables have
on LA-ICP-MS measurements. The use of standard additions was briefly studied. It was
concluded that this technique is useful, provided the particle sizes in the sample are
sufficiently small.
A compact, inexpensive laser was also evaluated for LA-ICP-MS measurements.
It was used for the analysis of glass and aluminum samples and provided low to sub-ppm
detection limits for the analytes studied.
Vlll

CHAPTER 1
INTRODUCTION
Developments in materials research, environmental science, industrial quality
control protocols, and numerous other endeavors have placed a high premium on the
development of accurate and sensitive analytical techniques. The presence of trace
element impurities in ceramics, for example, have been shown to affect the physical and
mechanical properties of these materials (1). In the electronics industry, the presence of
alkali and alkaline earth metals may cause corrosion and degradation of microelectronic
devices. In addition, the alpha emitters U and Th can cause errors in storage devices (2).
Therefore, the presence of these elements and their concentrations must be accurately
determined. Geologists need to determine absolute and relative concentrations of
elements in a variety of geological samples, in an effort to discover the principles
governing their distribution and migration (3). In addition, the ability to precisely
measure isotopic ratios of such elements as Li, Sr, Nd, Pb, and U is important for the
radiogenic dating of materials and providing an understanding of terrestrial processes (3,
4). The effects of heavy metals (e.g. Pb, Cr, Hg, and Cd) on human health are well
established and techniques for monitoring suspected areas of contamination, such as
soils, are required (5, 6). Furthermore, research has shown that the presence of particular
elements (e g. Se) results in some protection against heavy metal toxic effects; therefore,
it is important to be able to monitor the presence of several elements in the sample of
1

2
interest (7). It is evident that precise and accurate measurement of multiple elements at
low levels is required in a variety of applications.
In many cases the sample of interest is a solid and the ability to analyze the
material directly would be beneficial. Direct analysis of solids eliminates the need for the
time-consuming digestion and dissolution procedures required for many materials. In
addition, there is a reduced risk of sample contamination, incomplete digestion, or the
loss of some analytes during sample preparation. Lower absolute detection limits are also
possible since there is no dilution of the sample prior to analysis. In some instances, the
local concentration of analytes is of greater importance than the bulk concentration and
this information is completely lost with dissolution of the original sample. In these cases,
the ability to probe particular areas of the solid is essential and techniques capable of
microanalysis are needed.
Several of the techniques commonly used for the elemental analysis of solids are
X-ray fluorescence (XRF) (8), electron probe microanalysis (EPMA) (9), secondary ion
mass spectrometry (SIMS) (10), and glow discharge mass spectrometry (GDMS) (11).
Each of these techniques has its strengths and weaknesses. XRF is a well established
technique for the analysis of conducting and nonconducting solids; however, it suffers
from relatively poor detection limits (10-100 ppm) and significant matrix effects when
analyzing thick samples (3, 8). In addition, it is essentially a bulk technique and provides
poor spatial resolution. EPMA is a valuable surface analysis technique due to the high
spatial resolution (0.1-0.5 pm) that it provides. The technique is limited by high relative
detection limits (100-1000 ppm), sample charging with nonconducting samples, and the
requirement that the sample be held in vacuum. In addition, quantitative work requires a

3
highly polished surface and strict matrix-matching (9). SIMS is another technique for
surface analysis, but unlike EPMA, it offers very high sensitivity with detection limits as
low as 1 ppb (10). It is widely used for depth profiling studies because of its high in-
depth resolution (1-2 nm). Lateral resolution is typically around 0.1-1 jim (10). Like
EPMA, sample charging is a major problem when analyzing nonconducting samples and
the sample must be held in vacuum. In addition, matrix effects are very severe making
quantitation extremely difficult (12, 13).
GDMS is a technique which has been widely used for the analysis of conducting
samples. This is due to low detection limits (1-100 ppb) and high measurement precision
(11). Nonconducting samples can be analyzed by diluting the sample in a conductive
matrix (11), using a secondary cathode (14), or by operating the glow discharge in the rf
mode (15). The latter technique has shown considerable promise for the analysis of
nonconductors (e g. ceramics and glasses), but a major challenge to accurate quantitative
measurements is matrix effects based on sample thickness. The GD technique can be
used for depth profiling studies with a resolution approaching 100 nm. The lateral
resolution; however, is usually only a few millimeters (16). A major benefit associated
with GDMS, like SEMS, is that it provides isotopic information by virtue of its detection
methodology.
The use of laser ablation inductively coupled plasma mass spectrometry (LA-ICP-
MS) for the analysis of solid samples compares favorably with other techniques. A
focused, pulsed laser beam can be used to sample essentially any type of material and
when combined with the high sensitivity of ICP-MS, the technique offers sub-ppm
detection limits for most elements. Because the sampling and excitation steps are

4
separate, each can be independently controlled for optimum performance. This results in
reduced matrix effects compared to single step laser based techniques, such as laser
microprobe mass analysis (LAMMA) (17). Laser sampling also allows for modest
spatial profiling of the sample with lateral resolutions on the order of microns and in-
depth resolution on the sub-micron scale. Mass spectrometric detection also allows for
the measurement of isotopic ratios. The versatility of the technique is evidenced by the
many types of samples which have been analyzed, including metals, glasses, ceramics,
geological materials, biological materials, and polymers. One major advantage of LA-
ICP-MS compared to other solids techniques is that the sample can generally be analyzed
without any modification and is held at atmospheric pressure. This results in rapid
sample changing and increased throughput. In addition, elemental sensitivity factors for
LA-ICP-MS are relatively uniform for most elements when compared with many other
solids techniques. This makes the technique excellent for semiquantitative analyses (12);
however, quantitative measurements by LA-ICP-MS are often difficult to obtain.
The major difficulty associated with laser sampling is the complex nature of laser-
material interactions (18). This results in an unknown quantity of material being ablated
with a composition which may or may not accurately reflect the bulk sample. The
properties of the ablated material (e g. composition and particle size distribution), depend
on the laser and material characteristics. Because of the complex nature of the laser
sampling process, matrix-match standards are often required for quantitative
measurements on a given sample. This represents the major obstacle to the widespread
applicability of the technique, since for many solid matrices, suitable standards are not
readily available. To overcome these difficulties, approaches such as the use of fused

5
glass beads (19, 20) or pelletized mixtures of diluent and analyte (20, 21) have been
explored. Although useful in some cases, the sample preparation that is required
eliminates a major advantage associated with laser sampling and any spatial variations
cannot be assessed.
The intent of the present work is to develop methodology for the accurate and
precise determination of analytes in a variety of solid matrices, including glass, silicon
nitride ceramic, and soil, by LA-ICP-MS. Optimum experimental parameters were
determined and the effectiveness of solution-based calibration for the analysis of solids
was evaluated. In addition, the utility of glass standard reference materials for calibration
of ceramics was examined. As an alternative to internal standardization, mass
normalization by measuring light scattering from the ablated material was studied. Its
applicability for various sample types was assessed. A compact, inexpensive laser source
was evaluated for the analysis of glass, copper, and aluminum samples. Future research
directions are suggested based on these studies.

CHAPTER 2
BACKGROUND
The inductively coupled plasma (ICP) has developed into the dominant excitation
source for elemental analysis. Its initial development was as an excitation source for
atomic emission spectroscopy (AES); however, in the past decade it has been widely used
as an ion source for mass spectrometry. Numerous publications dealing with the
fundamentals of inductively coupled plasma-mass spectrometry (ICP-MS) and some
applications are available (22-27). The major benefits of ICP-MS, compared to ICP-
AES, are superior detection limits, spectral simplicity, and the ability to measure isotope
ratios. Detection limits for solutions are typically 100-1000 times better for ICP-MS,
with typical values for most elements in the 1-100 ppt range. Mass spectra of the
elements are simple and unique. The natural isotopic abundance spectral pattern provides
a simple means of identifying sample constituents. The ability to measure isotopic ratios
is important in geological and nuclear applications; however, it also allows one to assess
the existence and extent of interferences. The major components of a typical ICP-MS are
the ICP, an interface system, ion lenses, mass analyzer, and a detector. These will be
discussed in greater detail.
Inductively Coupled Plasmas
The ICP is a partially ionized gas maintained by energy coupled to it from a radio
frequency (RF) generator, typically operated at 27 or 40 MHz. Energy, usually between
1 and 2 kW, is coupled to the plasma with a 2-4 turn copper coil which acts as the
6

7
primary of an RF transformer. The plasma itself acts as the secondary. Argon (Ar) is
typically used as the plasma gas, although hydrogen-argon, nitrogen-argon, xenon-argon,
and helium plasmas have been studied as well (22). The plasma is generated inside and
at the end of a quartz torch, which consists of an assembly of quartz tubes. An
illustration of the ICP torch is given in Figure 2-1. The outer gas flow (coolant) is
typically 10-15 L min'1 and serves both to protect the tube walls from melting and as the
main plasma support gas. The middle annular flow (auxiliary or intermediate) is used to
keep the plasma from melting the central injector tube. Typical values for this flow are
0.5-1.0 L min'1. The inner gas flow (nebulizer or carrier) is used to puncture the plasma
and inject aerosol from the sample introduction system. This flow is typically in the
range of 0.7-1.0 L min'1.
The copper load coil is positioned around the end of the torch and supplied with
RF current. This produces a time varying magnetic field (27 or 40 MHz) which lies
along the torch axis. A Tesla coil is used to seed the Ar flow with free electrons. These
electrons precess around the magnetic field in circular orbits and the energy supplied to
the coil is converted into kinetic energy of the electrons. At atmospheric pressure, the
free electron path before colliding with Ar atoms is only around 10'3 mm; therefore, the
plasma is rapidly heated and a bright discharge is formed (23).
At the RF frequencies used, the skin effect is occurring which ensures that most
of the energy is coupled to the induction region of the plasma. The nebulizer gas punches
a channel through the center of the plasma and there is little mixing with the outer
annular region. Heat is transferred to this central gas flow mainly by radiation and
conduction from the annular induction region. Gas kinetic temperatures in the central

Sample Aerosol
and Nebulizer Gas
Ct
r.
Intermediate
Gas
n
ate
Coolant Gas
Figure 2-1. Inductively coupled plasma torch.

9
region are typically between 5000-7000 K, while the temperature in the outer region is as
high as 10,000 K (23). It is important that the power is coupled mainly into the outer
region where there is little interaction with the sample aerosol. This is because the
sample composition can vary significantly and have only minimal effects on the plasma
sustaining processes. Separation of the energy coupling and sample excitation region is
one of the major reasons why the ICP is characterized by relatively few chemical and
physical interferences.
The role of the plasma is to convert the sample into free ions. Most analyte ion
formation results from collisions of electrons, and the probability of this process is
dependent on the electron density (~1015 cm'3) and temperature (26). The gas
temperature, which describes the kinetic energy of Ar atoms, controls the desolvation and
vaporization of the sample aerosol. Transit times through the center of the plasma are
typically several milliseconds; therefore, desolvation, vaporization, atomization, and
ionization processes occur very efficiently. The degree of ionization is dependent on the
ionization conditions in the plasma, which are dominated by the major plasma constituent
(typically Ar, H, O, and electrons) and the partition function and ionization energy for the
atom of interest (23). The Saha equation, though not strictly valid since thermal
equilibrium does not exist in the plasma, is often used to give insight into ionization
conditions in the plasma. Under normal operating conditions, elements with ionization
energies of 8 eV or less (approximately half of the elements in the periodic table) are >90
% ionized in the plasma, while Ar with an ionization energy of 15.75 eV is only about
1% ionized (27). In addition, only a few elements (e.g. Ba, Sr, and Pr) have second

10
ionization energies low enough to produce a significant population of doubly charged
ions. The end result is a very efficient ion source for primarily singly charged ions.
Inductively Coupled Plasma-Mass Spectrometer Interface
The ability to extract ions representatively from the plasma is obviously quite
critical in ICP-MS. A description of the ICP-MS interface is provided with particular
emphasis on the interface used in this work (Figure 2-2). It consists of three cones
through which the pressure drops from atmosphere to approximately 10"5 torr by means
of differential pumping. The first cone, the sample cone, is located approximately 13 mm
from the top of the load coil and has an orifice of 1.1 mm. The region behind the sample
cone is maintained at 2-3 torr by a mechanical pump. This results in the formation of a
supersonic jet expansion. The jet consists of a freely expanding region, the zone of
silence, surrounded by shock waves called the barrel shock and Mach disc. Shock waves
result from collisions between the supersonic jet and residual gas in the expansion
chamber. To avoid losses of ions due to collisions and scattering, the second cone
(skimmer) is positioned before the Mach disc inside the zone of silence. Studies have
indicated that maximum ion transmission occurs when the sampler-skimmer distance is
approximately 2/3 of the distance to the onset of the Mach disc (28). The location of the
skimmer cone is 8 mm from the sample cone and the diameter orifice is 0.8 mm to allow
the centerline of the expansion to pass while removing the cooler gas at the edge of the
supersonic jet (29). Only about 1% of the gas that is sampled passes through the
skimmer cone. The pressure in the region behind the skimmer cone is maintained at 10'3
torr by a turbomolecular pump.

Sample Cone
Figure 2-2. Schematic of ICP-MS interface.

12
Unique to the Finnigan MAT SOLA used in this work is the incorporation of an
accelerator cone. It is located 10 mm behind the skimmer cone and has a +2 kV potential
applied to it. The lens is used to focus the ions that pass through it and accelerate them to
the ion optics. The purpose for this will be presented more clearly in the discussion on
ion optics.
Ion Optics
The purpose of the ion optics is to supply the quadrupole with ions of low enough
energy that they can be efficiently differentiated with respect to their mass to charge
ratios. Ions must be separated from neutrals, and photons produced from the plasma must
be removed since these can activate the detector and contribute significantly to the
background. Typically, a Bessel box has been employed for separating these
components. The Bessel Box consists of a central photon stop for removing photons,
while neutrals are removed by pumping of the chamber. Ions are electrostatically steered
around the photon stop and refocused into an exit slit. This arrangement results in two
detrimental effects; space charge effects and mass dependent focusing of the ion beam
(29). Space charge effects are more significant for lighter ions since they are more easily
deflected from their flight path. Mass dependent focusing results because different mass
ions have different kinetic energies and thus have different paths through the ion lenses.
Therefore, optimization of ion transmission is mass dependent and a compromise setting
must be used. This is typically in the mid mass range for multielement work, resulting in
a transmission profile which falls off at the low and high mass ends. The Finnigan MAT

13
SOLA ICP-MS accelerator cone and ion optics are designed to reduce these effects and
produce a more even transmission profile.
The accelerator cone supplies a high velocity stream of ions condensed in a
tightly focused beam. This beam is deflected off axis and realigned to exit parallel to the
original ion transmission, since ion losses of 50-80% have been measured when photon
stops are used (23). In addition, because the residence time in the ion optics is shorter
with the accelerated beam, space charge effects will be less significant. Since the beam
has not been defocused and refocused, as is the case with the Bessel Box, no mass
dependent transmission effects have been introduced. The final stage of the ion optics is
a phase matching lens that reduces ion velocities to the level required by the quadrupole.
Mass Analyzer
Several types of mass analyzers have been employed for ICP-MS, including
quadruples, magnetic sectors (30), time of flight (31), and ion traps (32). Only a
discussion of quadrupoles is provided here, since it was used in this work.
The quadrupole consists of four straight metal rods arranged parallel to and
equidistant to the ion axis. Opposite rods are connected together with DC and RF
voltages applied to each pair. The DC voltage is positive for one pair and negative for
the other. The RF voltages on each pair are 180° out of phase. Appropriate amplitudes
of these potentials produce trajectories for ions along the quadrupole which are stable for
only one particular mass to charge ratio. By keeping these potentials constant, single ion
monitoring is accomplished. To produce a mass scan, the amplitudes are varied while
keeping the RF to DC ratio constant. The mass transmitted is linearly related to the

14
magnitude of the RF and DC voltages. Peak jumping between masses of interest is
accomplished by rapidly selecting discrete values of RF and DC voltages. The use of
rapid and repetitive peak jumping is extremely important when dealing with transient or
highly fluctuating signals, such as those typical of LA-ICP-MS. The mass resolution is
determined by the ratio of RF and DC voltages. This is typically set to provide 1 amu
peaks across the mass range (5-240 amu).
Detection of Ions
Detection of ions is most often accomplished with a channeltron electron
multiplier. This detector is characterized by high gain (~108), low dark current, and fast
response time. Output pulses from the detector are sensed by a fast pre-amplifier and
sent to a discriminator and counting circuit. The detector is used for signal rates between
1 and 106 counts/second. For higher ion currents, a Faraday collector plate is used. This
detector can be used for ion signal rates between ~105 and 5 x 1010 counts/second;
however, the time constant of the DC amplifier limits the Faraday to relatively low scan
rates (64 ms or greater). In the Finnigan MAT SOLA, selection of the detector is
accomplished electronically by deflection of the ion beam to the appropriate detector.
Sample Introduction in ICP-MS
Samples are introduced to the ICP-MS as gas, vapor or a fine aerosol of solution,
or as solid particles. Pneumatic nebulization of solutions is the most common means of
sample introduction, due to its simplicity, sufficient stability, and an overall greater
understanding of the process. Although this dissertation is concerned with the

15
introduction of solid particles by laser ablation, mention is made of solution introduction
since it was often used in this work.
The principle behind pneumatic nebulizers is the disruption of a liquid stream by
a high velocity gas, resulting in the production of aerosol droplets. The liquid stream is
provided by a peristaltic pump which typically produces a 1 mL min'1 flow. In the
concentric type of nebulizer used in this work, liquid is introduced to a fine central
capillary and when the liquid reaches the tip, it is broken into fine droplets by the
shearing action of Ar gas flowing around the orifice at 0.5-1.0 L min'1. The liquid
particles produced can be as large as 100 pm; therefore, they must be further processed
before entering the ICP-MS. This is accomplished with a Scott-type double pass spray
chamber, which acts as a low pass filter. Larger particles settle out and only particles <10
pm are transported to the plasma.
The introduction of solutions, typically aqueous with small amounts of acid,
results in a large number of spectroscopic interferences. The dissociation of water
produces large amounts of oxygen and thus the presence of oxide species can be
significant. In addition, the constituents of acids used will also contribute to the
background spectrum. For this reason, nitric acid is often the preferred acid since its
constituents (H, N, and O) are already present in the plasma (23). Numerous methods
have been employed to decrease the solvent plasma load, such as cooled spray chambers
(33), Peltier effect coolers (34), membrane interfaces (35), and heater condensers (36).
Several techniques have been employed for introducing solids into the ICP-MS.
The uses and relative merits of these techniques are discussed in several reviews (22, 23,

16
27). They include slurry nebulization, electrothermal vaporization (ETV), direct sample
insertion (DSI), arc and spark ablation, and laser ablation (LA). Of these techniques, the
last (LA) is most promising since it can be used to sample a wide range of materials. In
addition to bulk analyses, it also allows for microanalysis by virtue of the focusing
characteristics of lasers. These attributes put laser ablation in a unique position among
solid sampling techniques for ICP-MS. The first application of LA-ICP-MS was reported
by Alan Gray in 1985 (37). Since that time, the technique has been used to analyze a
wide variety of materials, including metals (38, 39), glasses (19, 20, 40), ceramics (41,
42), geological materials (43, 44), biological materials (45), and polymers (46).
Laser Ablation
The ablation process (Figure 2-3) depends on a large number of factors, including
the nature of the solid material, laser characteristics (wavelength, pulse duration, etc.),
and pressure and composition of the gas medium. Nonetheless, a phenomenological
description can be given. When a short duration pulse of sufficient energy (typically
irradiances > 108 W cm '2) strikes the sample, the surface is instantaneously heated past
its vaporization temperature through linear one-photon absorption, multi-photon
absorption, dielectric breakdown, and additional undefined mechanisms (47, 48). Energy
dissipation of the surface layer is slow relative to the laser pulse duration and before this
layer can vaporize underlying layers have reached their vaporization temperature. The
underlying layers reach the critical point (temperature and pressure) resulting in an
explosion of the surface. This process is described as nonthermal and so no fractional
vaporization should occur (18). During the ablation process, a plasma is initiated on the
surface with temperatures in the range of 104-105 K (47, 48) and a duration of several

Target
Figure 2-3. Conceptual drawing of laser ablation.

18
microseconds. This also interacts with the sample surface and contributes to the quantity
and composition of the ablated material.
In addition to LA-ICP-MS, numerous other uses of laser ablation have been
explored. An excellent review by Darke and Tyson deals with fundamental aspects of
laser ablation and its uses in analytical spectrometry (17). These include techniques
which directly use the laser induced plasma as an excitation source (laser induced
breakdown spectroscopy - LIBS) (49), ionization source (laser microprobe mass
spectroscopy - LAMMA) (50), or atom reservoir (laser ablation laser excited atomic
fluorescence spectroscopy - LA-LEAFS) (51). As in LA-ICP-MS, laser ablation has
been used as a method of sample introduction for numerous analytical techniques
including flame and furnace atomic absorption (52, 53), glow discharge atomic emission
(54), and microwave induced plasma atomic emission (55). The major advantage of
separating the laser sampling step from spectroscopic detection is that it allows for
independent optimization of each step, potentially resulting in improved analytical
performance.

CHAPTER 3
GENERAL CONSIDERATIONS AND OPTIMIZATION
OF EXPERIMENTAL PARAMETERS
Introduction
The purpose of this chapter is to outline several of the key variables that affect
LA-ICP-MS measurements. It is intended to serve as a guide in the selection of
appropriate experimental conditions. The measurements have been performed on a
variety of samples (metals, glasses, ceramics, and soils); however, the goal is to provide
information that is generally applicable to all matrices. More specific information on
particular materials can be found in subsequent chapters.
Instrumentation
An illustration of the LA-ICP-MS setup is shown in Figure 3-1. It consists of a
Finnigan MAT SOLA ICP-MS (Hemel Hempstead, UK) and System 266 laser ablation
accessory. The ablation module consists of a Spectron SL 401 Nd:YAG laser which has
been frequency quadrupled to produce an output beam of 266 nm, as well as beam
steering and focusing optics and a CCD camera for remote viewing of the sample. The
CCD camera is mounted in parallel with the laser and aids in focusing of the laser beam.
The system also includes an x-y-z translation stage for adjusting the focus of the laser
beam and for selecting the area of the sample for analysis. The system is designed to
provide analysis only at a fixed location. However, it was modified to allow for
translation of the sample while the laser was firing. Laser repetition rates of up to 5 Hz
19

Figure 3-1. LA-ICP-MS System

21
can be utilized, with pulse energies of- 0.1-1 mJ and pulse widths around 15 ns in
duration. When focused to a 50 pm spot, laser irradiances (W/cm2) of - 8xl07-8xl08 are
achieved.
For analyses, the sample is placed in an ablation cell consisting of a 6 cm glass
tube with a quartz window for transmission of the UV laser beam. The cell has a total
volume of - 100 cm3. Ablated material is swept out of the ablation cell and to the ICP-
MS through 1.5 m of 3/16” i d. plastic tubing with a flow of argon. In addition, the
output from a concentric nebulizer and cooled Scott-type spray chamber can also be
continuously introduced to the ICP-MS during ablation with a glass Y-connector. This
dual sample introduction approach is termed “wet plasma” operation (38). A comparison
of wet and dry (ablation chamber flow only) plasmas will be made later in this chapter.
The ICP is typically operated between 1100 and 1300 W with coolant and intermediate
flows of 15 L/min and 0.9 L/min, respectively. Sample is carried to the plasma, whether
from the ablation chamber, nebulizer, or a combination of both, with a total argon flow of
-1 L/min.
Samples
Various samples were used in these studies. They included National Institute of
Standards and Technology (NIST) (Gaithersburg, MD) glasses, silicon nitride ceramic
bearings, various metals, and NIST soil samples. The soil samples were pressed into 1
cm pellets at a pressure of 5 MPa. No binder was required to produce relatively rigid
pellets. The only constraints on sample size were that they fit into the ablation chamber
and the height match the range of the stepper motor used to adjust the focusing lens
position. This stepper motor could be adjusted over a range of 1.25 cm and the lower

22
focusing position corresponded to a point ~0.2 cm above the base of the ablation cell.
Therefore, samples with thicknesses ranging from 0.2 cm to 1.45 cm could be analyzed
without any modification.
Results
Several experimental variables were investigated. These include plasma
operating conditions (wet and dry plasmas, RF power, and argon flow rate), laser
sampling mode (stationary or translated sample), and detection considerations (scan
parameters and internal standardization).
Plasma Operating Conditions
Among the attributes often cited for laser ablation as a sample introduction
technique is the elimination of solvents and hence, the resulting diminished level of many
background peaks. A comparison of background spectra from wet and dry plasmas
illustrated this point (Table 3-1). The dry plasma was operated at 950 W with a 1.0
L/min flow from the ablation chamber, while the wet plasma was operated at 1300 W
with a nebulizer flow rate of 0.65 L/min and an additional 0.35 L/min from the ablation
chamber. These operating conditions were chosen because they provided the highest
analyte signals for laser ablation measurements. In the wet plasma mode, a 2 % HNO3
solution was introduced to the nebulizer at 1 mL/min with a peristaltic pump (Gilson
Minipulse 3, Middleton, WI). Table 3-1 lists some of the major background peaks. In
each case, the background level is significantly larger for the wet plasma. In addition, the
introduction of solution can introduce trace levels of impurities even for high purity
reagents. The peak at m/z 58 was partially due to the Ni sampling cone.

23
Table 3-1. Comparison of wet and dry plasma background.
Background Species
(m/z)
Dry Plasma Signal
(counts/s)
Wet Plasma Signal
(counts/s)
CO+, N2+ (28)
3.3xl05
l.OxlO7
N2H+ (29)
1.9xl05
1.3xl06
NO+(30)
3.8xl06
3.0xl08
ArH2+ (42)
3.2xl04
2.4xl06
Ar>T (54)
2.5x10s
1.5xl07
ArO+(56)
1.3xl05
6.5xl06
ArO+, Ni+(58)
2300
6200

24
The effect of RF power on analyte signal for both dry (a) and wet (b) plasmas is
shown in Figure 3-2. The 63Cu signal was measured from laser ablated brass samples as
a function of RF power. As previously indicated, the wet plasma optimizes at higher RF
powers for the same total flow (1.0 L/min) through the torch injector tube. The reason
for this behavior was the energy required in vaporizing the solution aerosol. It should
also be mentioned that RF power optimization only applied to particular flow conditions,
since these variables were interactive. For example, if the flow was kept at 1 L/min and
the ratio of nebulizer gas to ablation chamber gas was increased, a higher RF power was
required for maximum analyte signals due to the increased solvent loading. For both wet
and dry plasmas, increasing the total flow through the injector tube required higher RF
powers for efficient ionization.
Optimized plasma conditions were more easily determined for the dry plasma
since only two variables (RF power and argon flow) were involved. The RF power was
adjustable over the range of 950-1450 W for stable plasma operation. With the wet
plasma, the nebulizer flow rate was more difficult to optimize. To gain insight into the
proper nebulizer and ablation chamber flow rates, the RF power was held constant at
1350 W while the nebulizer and ablation chamber flow rates were systematically varied.
Total argon flow rates between 0.8 and 1.2 L/min with individual flows from 0.2 and 1.0
L/min were studied. These studies indicated that ablation chamber flows of 0.3-0.4
L/min and nebulizer flows of 0.6-0.7 L/min were most suitable in terms of both
sensitivity and precision.

Cu Signal (counts/s)
Figure 3-2. Effect of RF power on Cu signal for (a) dry and (b) wet plasmas.
N>

26
Although background levels were higher, wet plasma operation was preferred for
most LA-ICP-MS measurements. This was because it provided a convenient means for
tuning the ion optics, gave better sensitivity and precision, and was useful in calibration
studies as discussed in Chapters 4, 5, and 7. Tuning of the ion optics in the dry plasma
mode involved continuously ablating a sample and adjusting the ion lenses to maximize
the signal from a matrix element. For multielement work, a mid-mass element was
chosen for tuning. Because the amount of material ablated by each laser shot varied and
different size particles were being introduced to the ICP-MS, laser ablation signals tended
to be noisier than those originating from nebulized solutions. The highly fluctuating
signal made tuning of the ion optics difficult.
With the wet plasma, solution-based analytes could be used for tuning. This
made tuning much easier since the signal was more stable. The major requirement for the
applicability of solution-based tuning was that analytes in the solution aerosol behaved
similar to those in laser ablated solid particles. Because of the high temperature of the
ICP and the constant plasma conditions (i.e. solution was continuously introduced
whether ablating or not), this appeared to be the case (Figure 3-3). These plots
demonstrated that analyte signals from laser ablated NIST 611 glass (Co, Ni, Cu, Sr, and
Ba) and a nebulized 20 ppb solution (Mg, Al, Mn, Co, Ni, Cu, Zn, Sr, and Ba) optimized
at similar RF powers. In addition, all analytes optimized at the same RF power. Similar
behavior was observed for soils, ceramics, and metallic alloys. This behavior was critical
for the use of solution-based calibration for the analysis of solids, as will be discussed in
Chapter 4.

Normalized Intensity
Figure 3-3. Effect of RF power on analyte signals for (a) solution and (b) laser ablated glass.
to
-j

28
Table 3-2. Comparison of sensitivity and precision for wet and dry plasmas.
NIST 611 glass
Sensitivity-Dry
Mn
Ni
Cu
Zn
Sr
(cps/ppm)
Sensitivity-Wet
26
130
49
10
41
(cps/ppm)
Precisiona-Dry
3180
1050
1590
170
1660
(% rsd)
Precisiona-Wet
4.8 %
16%
9.4 %
13%
13%
(%rsd)
4.9 %
8.9 %
7.1 %
5.6 %
5.9%
For precision measurements, analyte signals were normalized by the signal from Co.

29
The sensitivity and precision of LA-ICP-MS measurements was different for wet
and dry plasmas. Table 3-2 compares the sensitivity (signal/ppm) and precision (% rsd)
of laser ablation measurements for several elements in NIST 611 glass with both plasma
operating modes. The concentration of analytes was around 500 ppm in the glass sample.
The results indicated that much higher sensitivity was obtained with a wet plasma.
Several workers have studied the effects of water in the ICP and found that the
presence of some amount of water provided higher excitation than the complete absence
of water (56, 57). This is related to the presence of hydrogen produced from the
dissociation of water. The thermal conductivity of hydrogen is about ten times that of
argon, resulting in greater energy transfer in the ICP (56). There is obviously an
optimum amount of water that should be added to the plasma. The quantity of water
introduced to the plasma could be controlled by adjusting the temperature of the spray
chamber. When the spray chamber was cooled, water condensed on the walls, resulting
in smaller quantities of water being introduced to the plasma. The effect of spray
chamber temperature on analyte signals was studied and the results for Ba are shown in
Figure 3-4a. An increase in ion signals with decreasing spray chamber temperatures was
observed for all analytes studied (Be, Mg, Co, In, Ba, Pb, and U). In addition, the
decreased solvent loading resulted in lower oxide levels as shown for ArO (Figure 3-4b).
The level of oxides for Ba and U were also studied and showed a decrease of about a
factor of two between 25 °C and 0 °C. Unfortunately, some background species (Ar2 and
ArN) actually increased slightly with decreasing spray chamber temperatures.

Ba Signal (cps)
1.0x10^“
I
8.0x10s ”
6.0x10s-
4.0x10s
I
X
i
t ' 1 1 1 ' r
5 10 15 20
Q.
o
73
c
tí
c/3
9
<
3.6X106-
3.4X106-
3.2X106-
i
T
15
I
T
20
Spray Chamber Temperature (°C)
Spray Chamber Temperature (°C)
Figure 3-4. Effect of spray chamber temperature on (a) Ba and (b) ArO signals.

31
As indicated in Table 3-2, the precision of LA-ICP-MS measurements improved
with wet plasma operation. The reason for poorer precision in dry plasma measurements
was not entirely clear; however, it could be related to the different ablation chamber flow
rates. The lower flow rates associated with wet plasma conditions allowed for more
mixing of the ablation products from successive laser shots and produced more stable
signals. For dry plasma, less mixing occurred and signal fluctuations were greater in
magnitude. Because the quadrupole MS is a scanning instrument and spends a finite time
at each mass, highly fluctuating signals can lead to reduced precision when measuring
several isotopes. This will be illustrated more clearly in the consideration of scan times
later in this chapter.
Another potential benefit of the lower ablation chamber flow associated with wet
plasma operation was that it probably introduced smaller particles to the ICP. The
fraction of material transported through the transfer tubing as a function of flow was
estimated (Figure 3-5). The simple model (58) used in these calculations was based on
uncharged spherical silica particles and assumed that gravitational settling (rather than
convective diffusion) was the dominant loss mechanism. Arrowsmith and Hughes
studied the entrainment and transport of laser ablation products produced from Mo metal
(59). They determined that ~90 % of the products were entrained, ~40 % were
transported to the ICP, and concluded that the major loss mechanism was gravitational
settling. Because of differences in flow rates, tubing length, and ablation material,
different values would be expected for the present work. As demonstrated in Figure 3-5,
a significantly different particle size distribution would be transported to the ICP. For
example, 5 pm particles would not be transported to the ICP in the low flow case, but

Fraction Transported
32
Figure 3-5. Calculated transport efficiency for silica particles at
0.4 L/min. and 1.0 L/min. Ar flow rates.

33
nearly 50 % of these particles would be transported with the higher flow rates associated
with dry plasma operation. This is especially important considering that studies have
indicated that particles >3 jim are not completely vaporized and excited in the ICP (60,
61). Thepresence of these particles could lead to selective removal of more volatile
species in the ICP, resulting in inaccurate measurement of bulk concentrations. The
particle sizes produced from laser ablation are dependent on numerous factors, including
laser pulse duration, wavelength, irradiance, repetition rate, and physical properties of the
material. Studies have indicated that a wide range of particle sizes (tens of nanometers to
tens of micrometers) are produced during the ablation process (19, 21, 59).
Based on the comparison of wet and dry plasmas, all subsequent work was
performed in the wet plasma mode. Although the increased level of some interferences
was a drawback, the overall gain in analytical performance made it preferable.
Interferences were usually not a problem at most masses; however where they were,
another isotope of the element of interest was usually available. For example, 57Fe could
be measured instead of 56Fe, although with a loss in sensitivity due to differences in
natural abundance.
Scan Parameters and Internal Standardization
The scan parameters (number of channels, dwell time, and number of passes) and
data acquisition mode determined the total scan time for a measurement and played an
important role in the accuracy and precision of LA-ICP-MS measurements. The number
of channels refers to the number of discrete masses measured by the detector. In this
system, a total of 4096 channels are available. Dwell time refers to the amount of time
spent at each of these discrete locations (channels). The number of passes is simply the

34
number of times that the region or regions of interest are scanned over and averaged
during a measurement.
A procedure which combined the relative merits of both peak jumping and
scanning was used in the majority of this work. It involved scanning over a selected
region of interest (1 amu/peak) and then rapidly jumping to the next region of interest for
scanning, and so on. This allowed for the acquisition of true peaks, and therefore time
was spent only on regions of interest. Both peak height and peak area measurements
were used. Precision was similar for both types of measurements; but as expected, peak
area measurement produced lower detection limits.
The total measurement time for laser ablation analyses was typically around one
minute (300 shots @ 5 Hz). This was chosen to provide relatively rapid analyses, but
with enough laser shots and mass ablated to introduce a representative portion of the
sample to the ICP-MS. The amount of material ablated was dependent on the laser
irradiance and sample type, but was typically in the range of 2-50 ng per shot. It was
important to determine appropriate scan parameters for providing accurate and precise
measurements.
For these studies, NIST 611 glass was ablated and nine different isotopes (43Ca,
55Mn, 59Co, 60Ni, 63Cu, 88Sr,107Ag, 138Ba, and 165Ho) were measured on the multiplier
detector. For each isotope, 16 channels were measured with dwell times of 2, 8, 16, 32,
and 64 ms. The number of passes (128, 32, 16, 8, and 4) was adjusted so that the same
amount of time spent at each mass (~4 s). All measurements were normalized with an
internal standard (43Ca) to account for differences in mass ablated during and between
measurements. The average % rsd for the measurements was 3.8 % (2 ms), 3.5 % (8 ms),

35
5.8 % (16 ms), 7.4 % (32 ms), and 8.4 % (64 ms). The latter dwell time represented the
minimum dwell time that could be used with the faraday detector. These results clearly
indicated the benefit of using short dwell times and a large number of passes to average
out fluctuations in the laser ablation signal. Dwell times of 2 to 4 ms were used in all
subsequent multielement laser ablation measurements.
The use of an internal standard (43Ca in the above measurements) was important
for providing reasonable levels of precision. This was because a variable amount of
material was ablated due to differences in laser power (20 % rsd for n = 15 laser shots),
the nonlinearity of the ablation process (18), and the changing sample morphology (crater
formation). To account for these variations, the signal from a matrix element was
commonly measured. An important assumption was that the analytes and internal
standard were distributed the same over the sampled area, and that they exhibited similar
transport to, and excitation in the ICP. Internal standardization measurements typically
improved the precision by a factor of two or so. This improvement was greater in some
cases, for example, with soils and particulate samples. In these cases, the absolute signals
would routinely vary by a factor of three or more. In addition, when analyzing multiple
samples, it was difficult to ensure identical focusing of the laser beam with respect to the
sample surface. Internal standardization provided a convenient means of mass
normalization and eliminated the need to precisely control the position of the sample with
respect to the focused laser beam.
Sampling Strategy
Initial work with the LA-ICP-MS system was confined to single spot
measurements. The temporal profile of the 28Si signal from a glass sample obtained by

Si Signal (counts/s)
36
20 40
60
Time (s)
80 100
120
Figure 3-6. Temporal profile of Si signal from glass obtained
with repetitive pulsing at a single spot.

37
repetitively firing the laser at a fixed location is shown in Figure 3-6. The benefits of
repetitive pulsing, as opposed to single shot measurements, was that it allowed material
from successive ablations to mix. This provided a continuous laser ablation signal. The
MS could be repetitively scanned and information on the bulk sample was obtained.
Typically, the initial signal was higher and then fell off to a lower “steady-state” level
after 20 or 30 s. For this reason, a scan delay of 30 s was utilized in single spot
measurements to minimize any bias resulting from measuring signals in the initial region
where they changed very significantly, since the MS spent a finite time at each mass.
To determine whether the drop in signal was related to signal suppression or
simply differences in the ablation efficiency, integrated measurements of smaller
numbers of shots were performed. The laser was fired 50 times and the integrated 28Si
was measured. This was repeated in sets of 50 up to a total of 300 shots. Figure 3-7
indicates that the large initial signal was the result of a larger mass ablated rather than
signal suppression. In addition, the laser was focused both above and below the laser
surface to see if this had any effect on the observed behavior. Regardless of the laser
focus, a larger mass was initially ablated and then a relatively constant amount was
produced. This was not surprising since the first laser shots were incident on a flat
surface; whereas, all subsequent shots were directed at the crater formed in the sample.
To take advantage of the higher initial ablation yield, the instrumentation was
modified to allow for translation of the sample at 15 pm/s while the laser was repetitively
fired. Faster translation rates tended to produce more erratic signals and also required
that the direction of the stage be changed more often because of the limited sample width.
As expected, translation resulted in improved sensitivity because of the higher ablation

Silicon Signal (counts)
38
5x10 9 -i
4x10 « -
3x10 9 -
2x10 9 _
lxio 9 _
0 _
0
A 0.5 mm above surface
U 0.5 mm below surface
^ at surface
é
1 1 1 1 1 1 1 1 r
2 4 6 8 10
Set # (Each Represents 50 Shots)
Figure 3-7. Si signal as a function of shot number.

39
rate. The effect was greatest with pressed powder samples since the craters formed in the
surface were deeper, resulting in relatively rapid defocusing of the laser beam. A
comparison of craters produced from single spot and translation sampling is shown in
Figure 3-8. Measurements of the ablation depth and volume for these craters will be
discussed in Chapter 5. Another potential benefit of translation is that a larger area on the
surface can be sampled, minimizing the effects of local lateral heterogeneity. A
drawback is that a smaller surface layer is removed when translating and surface
contamination and/or heterogeneity can be problematic. This will be demonstrated with
glass samples in Chapter 4.

40
(a)
(b)
Figure 3-8. Laser produced craters in silicon nitride with
(a) single spot and (b) translation sampling.

CHAPTER 4
ANALYSIS OF GLASS SAMPLES
Introduction
Numerous studies on the analysis of glasses by laser ablation inductively coupled
plasma mass spectrometry (LA-ICP-MS) can be found in the literature. These studies
deal with a wide range of applications, from elemental fingerprinting of crime scene
evidence (62) to the use of glasses as calibration standards for geological materials (63-
66). In the glass industry, as the number of formulations and applications of glass
materials continues to grow there is an increased demand for rapid, accurate, and precise
determinations of the concentrations of elemental constituents (67). Direct methods of
analysis are preferred because of the difficulty in digesting these materials. Dissolution
generally consists of three steps: (1) treatment with HF; (2) further oxidation by addition
of HNO3, HCIO4, or H2O2; and (3) final addition of HC1 or HNO3 (68). When digesting
in an open vessel, silica is lost by formation of volatile SÍF4. An alternative to wet
chemistry is melting the sample with a suitable flux, such as sodium hydroxide or lithium
metaborate. The major problem with this method is the production of high salt
concentrations, which can lead to significant matrix effects and/or block the nebulizer
and ICP-MS sample cone (68). Because of the difficulties and time required for both
methods, LA-ICP-MS is an attractive alternative; LA eliminates the need for extensive
sample preparation and also offers the potential for information on the spatial distribution
of elemental constituents.
41

42
Glasses have been studied in this work for several reasons. They provided a
relatively homogeneous, analyte-rich matrix for characterization of the LA-ICP-MS
system. They were certified for several elements and therefore provided immediate
feedback on the suitability of non-matrix matched calibration strategies. As previously
mentioned, this was of particular interest because of the lack of matrix-matched standards
for many materials of interest (e.g. ceramics). The use of solution-based calibration for
the analysis of glasses will be addressed in this chapter and some of the important
findings discussed.
Experimental
The LA-ICP-MS system has been described previously (Chapter 3). Table 4-1
lists the typical ICP-MS operating conditions. The laser was operated at 5 Hz with
energies ranging from 0.1-0.7 mJ. Both single spot and line sampling were used.
NIST glass samples (611, 612, 614, and 617) were used in these studies. These
are synthetic Si, Na, Al, and Ca glasses which have been spiked with 61 different
elements at nominal concentrations of 500 ppm (NIST 611), 50 ppm (NIST 612), 1 ppm
(NIST 614), and 0.02 ppm (NIST 617). They are in the form of 1-3 mm thick discs.
For the solution-based calibration studies, a 10 ppm multielement standard (High
Purity Standards Charleston, SC) was diluted with deionized water and Optima-grade
HNO3 (Fisher Scientific, St. Louis, MO) was added to bring all solutions to 2 % HNO3.
Concentrations of the standards ranged from 1 to 50 ppb. A 2 % HNO3 blank was
continually introduced during laser ablation analyses via a glass Y-connector at the base
of the ICP torch. In this way, identical plasma conditions were maintained whether an
ablated solid or nebulized solution was being introduced.

Table 4-1. ICP-MS operating conditions.
Rf Power
1200 W
Coolant gas flow rate
15 L/min
Auxiliary gas flow rate
0.9 L/min
Nebulizer gas flow rate
0.6 L/min
Ablation chamber flow rate
0.4 L/min
Solution uptake rate
1.0 mL/min
Scan Conditions
Faraday Scans (major elements)
Scan range per isotope
1 amu
Number of passes
32
Number of channels per amu
8
Dwell time
64 ms
Multiplier Scans (minor elements)
Scan range per isotope
1 amu
Number of passes
128
Number of channels per amu
16
Dwell time
4 ms

44
The ion lenses, nebulizer gas, and RF power were optimized using the (115In)
signal from a 1 ppm solution (measured on Faraday). Both the Faraday and multiplier
detectors tuned at almost identical ion lens settings; therefore, either detector could be
used with the optimized settings. An ICP-MS response curve (Figure 4-1) was generated
for typical tuning conditions using a multielement solution. The absolute sensitivity
(number of ions detected / number of atoms introduced) was calculated for nine different
analytes (Be, Mg, Ni, Co, In, Ce, Pb, Bi, and U) which encompassed the mass range from
9 amu (Be) to 238 amu (U). The resulting plot indicated that the instrument was most
sensitive for mid-mass analytes and fell off for low and high mass analytes. It was
investigated whether tuning with a low (24Mg) or high mass isotope (208Pb) had any effect
on the instrumental response curve. No significant differences were observed.
Typical scan parameters are listed in Table 4-1. For most multielement analyses,
a total scan time of 60-80 s was used. This resulted in the averaging of 300-400 laser
shots per measurement. A 30 s delay was typically used before the start of data
acquisition as discussed in Chapter 3. Both peak height and peak area measurements
were used. Peak heights were measured directly using the SOLA instrument control
software. Peak area measurements were performed by importing the ASCII scan files
into a spreadsheet program for integration of the spectra. Typically, 8 channels were
integrated for each peak. Temporal profiles of analyte signals were generated by
sequentially and repetitively monitoring masses of interest. This feature was
incorporated into the SOLA software suite; however, due to problems in the program,
only a limited number of masses could be monitored.

Sensitivity
45
Figure 4-1. ICP-MS response curve generated by nebulization
of a 1 ppm multielement solution. Sensitivity refers
to the number of ions detected per number of atoms
introduced to the ICP-MS.

46
Calibration Procedure
The feasibility of using solution standards for the semiquantitive analysis of solid
materials has been studied previously (38, 67, 69-71). Hager (71) developed a model that
used response factors determined from solution nebulization and modified them based on
element-dependent volatilization efficiencies. This work reported accuracies of only +/-
50 % for aluminum, steel, and copper standards. The technique does not appear to be a
general method because ablation of nonmetals is likely to produce molecular species,
such as oxides and silicates. In addition, a large fraction of the ablated material
transported to the ICP is solid particles that have undergone solid-liquid-solid, rather than
solid-monatomic gas, phase changes (69). The use of dual-sample introduction appears
to be a more promising alternative for analyzing materials where suitable calibration
materials are not available. Its utility has been demonstrated for geological samples (38,
67, 70) and metals (69).
The calibration procedure used in this work involved measuring the intensity of a
single isotope for each element of interest (including an internal standard) in solution
standards and determining relative sensitivity factors (RSF’s). The RSF is defined in
equation (4-1), where E represents the element of interest and IS represents the internal
standard.
(4 1) psf(e IS) - Sensitivity(£) Intensity(E) ^ Concentradon(IS)
Sensitivity(IS) Concentration^) Intensity(IS)
If (a), identical plasma and mass spectrometer conditions prevail when nebulized
calibration solutions or laser ablated solid particulates are being introduced to the ICP-
MS (see Figure 3-2) and (b), the ablation material is an accurate representation of the
bulk material, then RSF’s obtained from solutions and solids will be equal, i. e.

47
(4-2) RSF(E/IS)soiution = RSF(E/IS)soiid-
Provided that the concentration of the internal standard is known in the solid of interest,
the RSF’s can be used to calculate the concentration of analytes from the expression
(4-3) Concentration^ = lnlensily(E)-‘ . C°ncentration(IS)„„,
Intensity(IS)
solid
RSF
solution
A major requirement of this solution-based methodology is that the concentration
of some element (the internal standard) must be known prior to analysis. This can either
be measured by some complementary technique, such as x-ray microanalysis (EPMA), or
determined from the stoichiometry of the solid. This is obviously a limitation of the
technique and a potentially useful method for overcoming this limitation is discussed in
Chapter 6. Typically, a minor isotope of a major matrix constituent is chosen as the
internal standard. In the analysis of these certified glass materials, 43Ca was found to be
the most suitable internal standard because of relatively low levels of interference at this
mass, and because signals could be measured with the multiplier detector. Signals
greater than 106 counts/s (cps) resulted in saturation of the electron multiplier detector,
which was exclusively used for the multielement analysis of trace elements in these
glasses.
Results
Fractionation
The production of ablated material that is an exact representation of the original
sample is essential in non-matrix-matched calibration of bulk materials. For
inhomogeneous samples, this requires that enough material is removed so that local
concentration variations in the solid are averaged out. For samples that are
homogeneously distributed on the scale of sampling, the major problem is fractionation

48
of species during the ablation process. Internal standardization can generally account for
differences in mass ablated both within and between matrices; however, fractional
ablation is more of a problem since fractionation causes enhancements or reductions in
relative analyte signals. Several workers have studied the role and extent of fractionation
in laser ablation measurements (69, 72-78). These investigations have revealed the
importance of laser wavelength, irradiance, and sample characteristics in determining the
extent of ablative fractionation. In addition, transport fractionation has been revealed as a
source of fractionation as well (75).
Two mechanisms have been proposed for ablative fractionation: volatilization of
non-refractory elements from sample areas outside the bulk ablation area; and zone
refinement, where elements within the molten zone surrounding the ablation crater
undergo selective migration into the laser ablation spot. Recent work has indicated that
the latter mechanism is the most likely source of fractionation, since the extent of
fractionation correlated with melting points of the element compounds in the sample (75,
78). If the former mechanism was dominant, the boiling point or vapor pressure of the
element compound would be critical.
The effect of laser irradiance and wavelength on elemental fractionation has been
studied as well (69, 78). Cromwell and Arrowsmith (69) reported that the Pb/Cu ratio
produced from a brass sample with UV laser ablation was one order of magnitude higher
with lower laser irradiances (< 1.3 x 108 W/cm2) than at higher irradiances (> 109
W/cm2). They attributed the increased fractionation at lower irradiances to the increased
volume of molten sample relative to the bulk ablation volume. Recently, Figg and Kahr
(78) studied the effect of laser irradiance, as well as laser wavelength, on ablative

49
fractionation in glass samples. They reported much higher fractionation in the case of
ablation with 1064 nm and 532 nm wavelengths, compared to 266 nm. This agrees with
earlier studies that compared ablation behavior for IR and UV lasers (76, 77).
In the case of UV lasers, the primary means of material removal is by direct laser
interaction with the surface for the duration of the laser pulse. The initiation of the laser
induced plasma occurs in less than 1 ns; however, the UV beam is not significantly
absorbed and the role of the plasma in material removal is insignificant (77). With an IR
laser, the direct interaction only lasts a fraction of the laser pulse duration since the
plasma is highly absorbing at this wavelength. The plasma is heated by the laser energy
and the primary means of material removal is by the plasma-material interaction (77).
This produces local heating of the sample, resulting in greater fractionation of more
volatile species.
In this work, the effect of laser irradiance on relative analyte signals was studied
to determine if fractionation was occurring, and if so, to what extent. This was essential
if the goal of accurate quantitative measurements was to be realized. In this study, NIST
611 glass was repetitively ablated at 5 Hz with laser pulse energies of ~ 0.5 mJ. The
sample was not moved, since this would maximize local heating of the sample and induce
fractionation. The focus of the laser was adjusted over a range of 2 mm to adjust the
irradiance at the sample surface. The laser spot size, as determined by the surface craters,
ranged from 40 pm to 200 pm. Signals for eleven elements (nB, 55Mn, 59Co, 85Rb, 88Sr,
89Y, 90Zr, l33Cs, 182W, 208Pb, and 209Bi) were acquired over ~ 500 laser shots. The ratios
of intensities to the 55Mn intensity were determined to see if any changes were observed
over the range of irradiances. Mn was chosen because it’s oxide melting point (oxides

50
should be dominant species in glass) was intermediate for those elements studied. Table
4-2 lists the oxide melting points for several elements in the glass sample.
The results for these measurements are given in Figure 4-2. The intensity ratios
have been normalized to their value at the highest laser irradiance (~ 5xl08 W/cm2).
Therefore, the relative change in composition can be determined by the deviation from a
value of one. In Figure 4-2 (a), elements with oxide melting points lower than Mn are
plotted and in (b), elements with higher oxide melting points are plotted. Based on these
results, significant differences in the composition of the ablated material were not
observed for the low melting point elements until very low irradiance values (< 5xl07
W/cm ). Below this value, fractionation of more volatile elements occurred (as much as
2 times in the case of Bi). Since an irradiance of 5x107 W/cm2 was near the ablation
threshold, visual observation of the laser-induced plasma could be used as a guide to
ensure that the laser irradiance exceeded this value. The presence of a visible laser spark
on the sample surface could then be used as a guide to ensure representative sampling of
these elements in glass.
The behavior of those elements with higher oxide melting points was more of a
problem. The relative intensity of Co did not deviate significantly over the range of
irradiances studied; however, this was not surprising considering the proximity of the
oxide melting points for Co and Mn. The relative intensities of Y and Zr decreased for
irradiances below ~1.5xl08 W/cm2, indicating that these more refractory oxides were not
being ablated representatively. For some unknown reason, this trend was reversed at
lower laser irradiances. Nonetheless, these studies revealed that at laser irradiances of
2x10 W/cm and greater, a reproducible ablation composition was produced. This was

Table 4-2. Oxide melting points.
Element
Oxide Melting Point (°C)a
Bi
180,825
Ag
230
Pb
290-500, 886
Cs
400
Rb
400, 570
B
450
W
800-900, 1473, 1500-1600
Co
895, 1795
Cu
1235, 1326
Mn
1564
Zn
1975
Ni
1984
La
2307
Y
2410
Sr
2430
Ca
2614
Zr
2700
a from reference 79

Intensity Relative to Mn
Laser Irradiance (W/cm2) Laser Irradiance (W/cm2)
Figure 4-2. Effect of laser irradiance on relative analyte signals for
elements with (a) lower oxide melting points and (b) higher
oxide melting points than Mn.
LAl
K>

53
important considering that slight changes in the laser focus, resulting from crater
formation, surface roughness, or the analysis of samples with small differences in height
would not significantly affect LA-ICP-MS results. The case is different when spatial
information is sought, since the crater diameter and depth are dictated by the laser
irradiance. In this instance, ablative fractionation would be more significant and proper
consideration would have to be given to account for this effect if analyte concentrations
were sought.
The consistency of the ablation product did not guarantee that it was truly
representative of the bulk material. Some melting of the glass material did occur as
evidenced in Figure 4-3, which compares the unablated glass surface (a) to the inside of a
laser produced crater (b). This and all other scanning electron micrographs (SEM’s) were
taken with a JEOL 35CF electron microscope. The representativeness of the ablation
product would be assessed from solution calibration results.
Solution-Based Calibration
In the discussion on the solution calibration procedure, it was stated that accurate
results could only be obtained if two criteria were met: (1) identical plasma conditions
should be maintained for both solution and ablated solid analysis and (2) the ablation
process should produce a representative subsample. The latter aspect was recently
discussed. The criterion of identical plasma conditions was speculated to exist based on
similar RF optimization behavior for ablation and solution measurements (see Figure 3-
2). This point was more clearly illustrated when RSF’s obtained from a 10 ppb
multielement solution and laser ablated NIST 611 glass with the ICP-MS operated in

54
Figure 4-3. Comparison of (a) unablated and (b) ablated glass surface.

55
Table 4-3. Comparison of RSF’s for solution and laser ablation measurements.
Element
Solution RSFa
(sd)b
LA-Wet RSFa
(sd)b
LA-Dry RSFa
(sd)b
Mn
1.00 (.03)
0.87 (.04)
0.41 (.02)
Ni
0.20 (.01)
0.19 (.02)
3.1 (.5)
Cu
0.51 (.01)
0.44 (.04)
0.46 (.04)
Zn
0.064 (.005)
0.046 (.003)
0.14 (.02)
Sr
0.607 (.02)
0.45 (.03)
16(2)
a relative to Co
b standard deviation for n=5 measurements

56
both the dry and wet plasma (dual sample introduction) modes were compared (Table 4-
3). All results were relative to Co, which was present in the glass at around 390 ppm.
The RSF’s obtained from solution and those for laser ablation with dual sample
introduction mode agreed very well with one another (5-28 %), especially considering
that a noncertified minor element was being used as the internal standard. This indicated
that under the experimental conditions used, the glass matrix was representatively
sampled. Laser ablation measurements with a dry plasma resulted in considerably
different RSF’s compared to solutions (10-1450 %), even though identical laser sampling
conditions were employed. The discrimination, therefore, arose in the ICP-MS and
indicated the importance of matched plasma conditions for obtaining accurate results.
The need to measure an internal standard was somewhat restrictive; however, it
was extremely beneficial in providing stable and reproducible measurements. This was
not only in terms of correcting for differences in mass ablated, but also in accounting for
instrumental drift. This point is clearly illustrated in Figure 4-4. In this plot, the average
intensities of 59Co and 60Ni from a 50 ppb solution were plotted over a series of fifteen
scans (1 min/scan). Even though the solution was being continuously introduced at the
same rate, a significant decrease in signal levels was observed. It was later determined
that this was due to an aging detector. The use of absolute intensities would have been
meaningless in this case, but the relative intensities remained stable (1.2 % rsd) since all
isotopes exhibited the same behavior.
Similarly, it was observed that RSF’s obtained from solutions did not change
significantly over time due to similar ICP-MS tuning conditions. This was somewhat

o*
o
oo
§
J
1.2x10 6.
6.0x10
4.0x10
5_
5_
2.0x10
Co
1.0]
O
0.9:
â– 
Ü
os:
1.0x10
o
0.7:
â– 
i
'5
0.6:
0.5:
0.4:
â– 
c
0)
£
0.3:
0.2
8.0x10 5
# Ni
â– 
â– 
#
â– 
â– 
Ni/Co rsd = 1.2 %
“1 r—i 1 1-
2 4 6 8 io 12 14 16
Scan Number
$
T
10
~r
12
T~
14
16
Scan Number
Figure 4-4. Ni and Co intensities as a function of scan number (time).
Inset shows the ratio of the two intensities.

58
surprising, even though the same tuning procedure was used each time the instrument
was run. The implication was that RSF’s measured on one day were reliable for
considerable periods of time if no major instrumental modifications were made. To
illustrate this point, the element concentration values from a NIST 611 glass determined
by the solution calibration procedure were compared for three different days. The
solution RSF’s determined on the first day were used to calculate analyte concentrations
(see Equation 4-3) on that day, the next day, and one month later. The results are given
in Table 4-4 and indicate that significant differences in accuracy were not observed over
this period of time. The precision (% rsd, n=5 replicates) of the measurements was
typically 10 % or better. Considering that a trace element (Sr) was used as the internal
standard, the accuracy and precision of these measurements was quite good. In later
measurements, an isotope of a major matrix constitiuent (43Ca) was used as the internal
standard. For all of these measurements, ablation and data collection were performed at
five different fixed locations on the glass surface. After this time, the instrumentation
was modified to allow for translation of the sample during ablation. A comparison of the
results obtained for these two different sampling modes will be presented.
Spot Sampling Versus Line Sampling
Sample translation during ablation was investigated because of several potential
advantages. Sample translation should have provided higher signals since a partially
fresh surface was being ablated by each laser shot and would be expected to give more
representative sampling for inhomogeneous materials. In addition, reduced fractionation
should have resulted since the laser was not defocused from crater formation and

59
Table 4-4. Results for NIST 611 glass based on solution calibration.
Cert. Conc.a
(ppm)
Meas. Cone,
(ppm)
Day 1
(% difference)
Meas. Cone,
(ppm)
Day 2
(% difference)
Meas. Cone,
(ppm)
Day 30
(% difference)
Mn
485
470 (-3.1 %)
454 (-6.3 %)
354 (-27%)
Ni
458.7
489 (6.6 %)
438 (-4.4 %)
339 (-26 %)
Co
(390)
389 (-0.2 %)
447(14%)
312 (-20%)
Cu
(444)
366 (-17%)
350 (-21 %)
342 (-23 %)
Zn
(433)
312 (-28%)
295 (-32 %)
269 (-37 %)
Rb
425.7
399 (-6.3 %)
413 (-3.0%)
434(1.9%)
T1
(61.8)
50.0 (-19%)
66.5 (7.6 %)
72.1 (17%)
Pb
426
366 (-14%)
415 (-2.6%)
464 (9.0 %)
U
461.5
352 (-24%)
445 (-3.7 %)
497 (7.8%)
Values in parentheses are not cerified

60
localized sample heating would be less significant since the same area was not
continuously ablated with the high powered laser.
For a typical 1 minute analysis (300 laser shots), translation of the sample at 15
pm/s produced signals that were on average 2 to 3 times higher than those obtained from
a single spot. The dimensions of a typical single spot and translated sample crater were
measured with an optical microscope. For single spot sampling, the crater was estimated
to be 75 pm wide and 120 pm deep. If a parabolic crater was assumed, this corresponded
to a total ablated mass of 830 ng (2.2 ng/shot). For translation, the crater was 50 pm
wide, 15 pm deep, and 900 pm long (assumed from 15 pm/s x 60 s). If a parabolic
trough was assumed, this corresponded to a total ablated mass of 1350 ng (3.8 ng/shot).
This provided further evidence that translation resulted in higher ablation efficiencies.
The exact magnitude of signal enhancement could not be inferred from these
measurements, since the sample was typically ablated for 100 shots (20 s) before signals
were acquired. It was previously shown (Figure 3-8) that the ablation efficiency
decreased significantly after the first 50 shots when single spot sampling was used.
NIST 611 glass was analyzed to assess the homogeneity of the material and to
determine if sampling strategy had any effect on the accuracy of solution calibration
based measurements. For this study, scans were made at different spots (25 total) on the
sample or at different lines (5 scans/line) produced from translating the sample. Several
isotopes (43Ca, 55Mn, 59Co, 60Ni, 63Cu, 88Sr, 208Pb) were measured for a total acquisition
time of one minute. Ca served as the internal standard in these measurements, since it
was a major matrix constituent (12 % CaO). The RSF’s determined from both sampling
methods were compared with RSF’s obtained from a solution containing 20 ppb of all the

61
elements of interest, except for Ca, which was present at 2 ppm. This closely matched
the ratios of trace element to Ca in the glass sample. The results for Mn, Co, Cu, and Sr
are presented in Figure 4-5. The translation values were significantly higher for Mn and
Co, slightly higher (statistically significant at 95 % level) for Cu and Ni (not shown), and
almost identical for Sr and Pb (not shown). Where differences did occur, spot sampling
measurements were more accurate as evidenced by their proximity to the solution results.
The most likely reason for these discrepancies was either inhomogeneity in the sample or
the presence of fractionation.
Examination of the glass surface with scanning electron microscopy revealed the
presence of a significant amount of redeposited particles (Figure 4-6). For comparison,
the top of this picture contains areas that had been previously ablated. After ablation, the
sample surface was wiped clean to remove any redeposited material. It was thought that
these redeposited particles might possibly be enriched or depleted in certain elements and
could influence the laser based RSF’s, making them vary more significantly compared to
the solution value. The effect would be more significant for translation analyses since
areas where particles were redeposited would be continuously sampled as the laser
probed the surface. Also, because the surface layer sampled was smaller with translation
the proportion of redeposited material to fresh material would be greater in this case. In
order to investigate these potential sources of variation, a larger set of elements was
studied and sampling was performed in a way that would maximize any effects due to
sampling of redeposited material.
In this study, elements encompassing a wide range of oxide melting points (see
Table 4-2) were measured since it has been shown that the extent of fractionation

RSF (Co/Ca) RSF (Mn/Ca)
62
Scan Number
Scan Number
Figure 4-5. Comparison of results for spot and line sampling for Mn, Co, Cu, and Sr.

RSF (Sr/Ca) RSF (Cu/Ca)
63
Scan Number
Figure 4-5 continued.

64
Figure 4-6. SEM of glass surface after ablation.

65
correlated with this property (75, 78). The isotopes used were 43Ca, 55Mn, 59Co, 60Ni,
63Cu, 66Zn, 88Sr, 90Zr, 107Ag, 139La, and 208Pb. The oxide melting points ranged from 230
°C for Ag to 2700 °C for Zr. Both single spot and translation measurements were
performed as previously described. In order to maximize the proportion of redeposited
material sampled, additional analyses were performed close to previous ablation tracks at
the same laser energy (0.5 mJ) and with lower energies (0.1 mJ). The use of lower laser
energies should result in the sampling of an even smaller surface layer. To ensure that
any differences observed at low energies were due to the sampling of redeposited
particles, analyses were also performed at a location well removed from any previous
ablation craters.
Table 4-5 summarizes the results obtained for glass ablated with single spot
sampling, translation sampling, and translation sampling through areas with significant
amounts of redeposited material. All of these measurements were performed with a laser
energy of 0.5 mJ. Significant differences (> 10 %) between spot and translation sampling
were observed for Mn, Co, Zr, and La. A statistically significant difference (at 95 %
level) was also observed for Ag. RSF’s for Mn and Co increased with line sampling,
while those for Zr, La, and Ag decreased. No major difference between translation
measurements was observed in this study, indicating that the particles did not have any
effect on the results at this laser energy. The differences observed for spot and translation
sampling could not be explained on the basis of fractionation. If fractionation was
occuring, it should have been more prevalent with spot sampling because localized
heating would be more likely in this case. Since Ag and Zr had the lowest and highest
oxide melting points, respectively, they should have displayed opposite behaviors. The

66
Table 4-5. RSF’s obtained from various sampling strategies.
Single Spot
RSF* (sd)b
Translation
RSF“ (sd)b
T ranslation-redep.
RSF“ (sd)b
Mn
530(20)
610(30)
620 (30)
Co
510(20)
580 (30)
580 (30)
Ni
122 (5)
121 (5)
123 (5)
Cu
230 (9)
240(10)
250 (10)
Zn
56 (3)
53 (3)
54 (3)
Sr
217(5)
218 (8)
216(7)
Zr
139 (5)
113 (5)
113(2)
Ag
120 (5)
112(4)
115(5)
La
101 (6)
89 (3)
84 (4)
Pb
16(1)
1-8 02)
1-8(2)
a relative to 4JCa
b standard deviation (n = 15 measurements)

67
RSF for Ag should have increased for spot sampling (observed), while Zr should have
remained stationary or decreased slightly (opposite observed). In addition, Mn and Co
exhibited the largest RSF changes, even though these elements did not possess melting
points significantly different from elements (e. g. Cu, Zn, and Ni) where no changes in
RSF’s were observed. The differences observed were most likely the result of small-
scale inhomogeneity in the glass samples. Measurements performed with a lower laser
energy supported this notion.
A comparison of RSF’s (Figure 4-7) obtained for spot sampling with a laser
energy of 0.5 mJ and translation sampling with energies of 0.5 mJ and 0.1 mJ revealed
significant differences. In this plot, RSF values were normalized to solution RSF values;
therefore, the accuracy of the measurements was readily apparent. For example, a value
of 1.1 indicated a 10 % error, 1.2 a 20 % error, and so on. The low energy measurements
were performed both near a previous ablation track and well beyond any previously
ablated areas. No significant difference between these measurements was observed,
providing more proof that the particles did not affect the accuracy of measurements. For
Ag and Cu, these low energy measurements resulted in significantly higher RSF’s
compared to higher energy measurements. This was almost certainly due to
inhomogeneity since an even smaller surface layer was being sampled in these cases. It
could have resulted from surface contamination, but this was unlikely since only these
two elements demonstrated this behavior. Likewise, if fractionation were at fault
differences would have been observed for other elements as well.
The results clearly indicated the importance of introducing a representative
portion of the sample during analysis. Single spot sampling resulted in better accuracy

RSF (X/Ca) Normalized to Solution
Figure 4-7. Comparison of RSF’s obtained for spot sampling (0.5 mJ) and line sampling (0.5 mJ and 0.1 mJ).

69
(average of 9.5 %) than translation mesurements (average of 14 %) for glass, since it was
less affected by in-depth inhomogeneity. Sampling depths were around ten times higher
with the former. The precision (% rsd) of the measurements was less than 5 % for both
sampling modes. Lateral inhomogeneity was not a problem in either case, due to the
relatively large crater diameters (50 pm or greater). This behavior should not be
generalized since the sampled volumes, homogeneity, and physical characteristics of the
sample all play a role in determining the most suitable sampling strategy.
Conclusions
Calibration using standard solutions and dual sample introduction was shown to
provide reasonably good accuracy (+/- 10 %) for trace elements in NIST glass samples.
Detection limits (3a) were less than 1 ppm for all elements studied. The effects of
ablative fractionation were studied and it was determined that at irradiances > 2xl08
W/cm2, representative sampling was achieved. Sampling strategies were compared and it
was determined that spot sampling produced more accurate measurements for the glass
samples. Translating the sample produced higher signal intensities; however, local
heterogeneity was more problematic because a smaller surface layer was sampled in this
mode. Ablated mass was estimated to be a couple of ng’s per shot.

CHAPTER 5
ANALYSIS OF SILICON NITRIDE CERAMIC BEARINGS
Introduction
There is considerable interest in using silicon nitride (SÍ3N4) bearings for a wide
variety of applications. Such interest results from the unique chemical and physical
properties which these ceramics possess. In comparison to conventional steel bearings,
silicon nitride bearings offer high speed and acceleration capability because of their low
density, extended temperature capability, longer lifetimes and lower wear rates, excellent
corrosion resistance, and the ability to operate under conditions of marginal lubrication
(80). This combination of qualities has led to investigations of silicon nitride bearings for
use in high speed, high temperature applications, such as in the aerospace industry.
The physical and mechanical properties of ceramic materials are influenced by
trace element impurities (81-84). Therefore, comprehensive trace element
characterization, in terms of both the bulk composition and spatial distribution of
elements, is required for ceramic materials if they are to be used in demanding
environments. Digestion procedures have been used for ceramics with analysis of the
resulting solution by several techniques, including ICP-AES/MS (85, 86) and Flame AAS
(85). These procedures are difficult and time consuming, due to the resistance of ceramic
materials to chemical attack. Wet chemical procedures can also introduce contaminants,
and they only provide information on the bulk sample and tell nothing of the distribution
of elemental constituents in the ceramic.
70

71
Direct sampling methods are preferred for the analysis of compact ceramics.
Several reviews discuss the methods currently used for analyzing ceramic materials (81,
82). Characterization of ceramics on both the bulk and micro scale is required in many
cases. Most often, several complementary techniques are necessary to provide this
information. LA-ICP-MS is extremely well suited for the bulk characterization of
ceramic materials, and also possesses the capability of providing spatial information on
the pm scale. The suitability of laser ablation results from the ability to sample
essentially any type of material with little to no sample preparation. The major obstacle
to the use of LA-ICP-MS for the quantitative analysis of ceramic materials has been the
lack of suitable standard reference materials. In some cases, synthetic ceramic standards
were created for analyzing ceramics with LA-ICP-MS (87, 88); however, these were
designed to analyze a specific material. It would be valuable to develop methodology
that would allow for the accurate analysis of a wide range of ceramics. This issue was
the driving force behind the investigation of solutions for calibration of solid materials, as
discussed in Chapter 4.
In this chapter, use of the solution calibration methodology for reliable
quantitative elemental analysis of silicon nitride ceramic bearings will be presented. In
addition, the use of NIST glasses for calibration of silicon nitride ceramics will be
discussed. The results obtained from the two methods were compared to one another,
and with results obtained by electron probe microanalysis (EPMA). Ablation craters
have been characterized using both profilometry and scanning electron microscopy. The
detection efficiency of the LA-ICP-MS system was estimated based on the profilometry
results.

72
Experimental
The LA-ICP-MS system (Chapter 3) and typical operating conditions (Chapter 4)
were previously discussed. The laser was operated at 5 Hz, with pulse energies of 0.7
mJ. Both single spot and translation sampling were used in this work. Typical analysis
times were 60 to 80 s (signals averaged over 300 to 400 laser shots).
Silicon nitride bearings (NBD-100 Cerbec, East Granby, CT) were mounted in
epoxy and cut with a diamond blade wafering saw to obtain a flat surface. They were
then polished to a 1 |im finish. These sample preparation steps were required for the
EPMA analysis only. Laser ablation analyses could be performed directly on the intact
bearing.
The solution calibration methodology was presented in Chapter 4. For the
analysis of silicon nitride, it would have been preferred to use a minor isotope of silicon
( Si or Si) as the internal standard directly; however, large interferences at these
masses resulted in saturation of the multiplier detector. Magnesium was chosen as an
alternative, since it was present at significant levels (1000s of ppm) in the samples, and
also possessed an isotopic pattern which allowed for its measurement using both the
Faraday (24Mg - 79 % abundance) and multiplier (25Mg - 10 % abundance) detectors.
The analytical procedure involved measuring 28Si and 24Mg with the Faraday detector and
determining the concentration of magnesium in the sample based on the solution
RSF(24Mg/28Si). The concentration of silicon (60 %) was estimated from sample
stoichiometry. All other elements (minor and trace) were measured with the multiplier
detector and the concentrations were determined from sensitivity factors relative to 25Mg.

73
In the glass-based calibration work, a National Institute of Standards and
Technology glass was used (NIST 611). Cobalt was used as the internal standard for
these studies, since its concentration in the glass was known and magnesium’s was not.
The calibration procedure was similar; RSF’s (analyte/59Co) obtained from the glass
matrix were used to calculate the concentrations of analytes in the silicon nitride bearing.
Ablation craters were characterized using an Alpha-Step 500 profilometer
(Tencor Instruments, Santa Clara, CA) and a JEOL 35CF scanning electron microscope
(SEM). For the SEM’s, the silicon nitride samples were attached to mounts with carbon
paint and coated with AuPd to provide a conductive surface. EPMA of the silicon nitride
bearings was used for comparison with LA-ICP-MS results. For these analyses, a carbon
coating was used since this resulted in lower background levels. The work was
performed on a JEOL Superprobe 733.
Results
Identification and Distribution of Elements in Silicon Nitride
A complete mass scan of the silicon nitride bearings (NBD 100) used in this study
resulted in the identification of 26 elements (Mg, Al, Ca, Sc, Ti, Cr, Ni, Co, Cu, Sr, Y, Zr,
Mo, Nb, Ag, Cd, Sn, Ba, La, Ce, Pr, Nd, Ta, W, Pb, and Bi). Although no attempt was
made to quantify all of these elements, their estimated concentrations ranged from the
hundreds of ppb level to thousands of ppm. For comparison, a newer grade of bearing
material (NBD 200) was analyzed and contained much lower levels of impurities, with
only the sintering aid (Mg) having a high concentration (> 100 ppm). Only ten elements
were present at high enough levels to be detected with the LA-ICP-MS system. For this

74
reason, the NBD 100 bearings were chosen for all subsequent work since they provided a
more analyte-rich sample.
For LA-ICP-MS to be used for bulk analysis of the ceramic bearings, it was
essential that the small surface layer sampled (~ 5 pm) on the exterior of the bearing be
representative of the bulk material (Figure 5-1). This was tested by analyzing the interior
and exterior surfaces of a cut bearing. The measured concentrations of all analytes were
almost identical within the measurement uncertainty, indicating that the relatively small
volume sampled by laser ablation of the surface provided information on the bulk
specimen. The SEM (Figure 5-1) revealed that the ablation process resulted in both
melting and fracturing of the silicon nitride surface. Melting was evidenced by the
presence of redeposited spherical particles around the ablation craters, as well as by a
comparison of the unablated (Figure 5-2 (a)) and ablated (Figure 5-2 (b)) silicon nitride
surface. A discussion of the particles produced, and their origins, from laser ablation of
ceramic, glass, and soil samples will be presented in Chapter 6. Fracturing of the silicon
nitride surface was evidenced by the small step-like features outside of the main ablation
track.
As mentioned in the experimental section, an isotope of magnesium was chosen
as the internal standard for most of the quantitative measurements. A requirement for the
use of internal standardization for obtaining accurate results is that the analyte and
internal standard be distributed similarly over the area sampled, and that they exhibit
similar behavior in transport to and excitation in the ICP. This was tested, albeit on a
limited basis, by measuring the temporal response of the 25Mg and 63Cu signal over a
period of 90 s, while the laser was repetitively fired (Figure 5-3). The two signals

75
Figure 5-1. Scanning electron micrograph of a laser ablation track in
silicon nitride. The sample was translated at 15 pm/s.
SEM was taken with the sample tilted at 50 degrees.

76
Figure 5-2. Scanning electron micrographs of (a) unablated and (b)
ablated silicon nitride surface.

77
Time (s)
Figure 5-3. Temporal profiles of Mg and Cu signals from silicon nitride.

78
correlated very well with one another, and partially justified the use of 25Mg as the
internal standard for quantitative measurements of trace constituents in the ceramic.
Additionally, the distribution of magnesium and titanium was studied in the silicon
nitride bearing. Signals for these two elements were measured from 24 different
locations on the sample. The % rsd of the elemental ratios (Ti/Mg) was 10 %, indicating
that titanium and magnesium were distributed similarly over the surface of the bearing.
Results from Solution- and Glass-Based Calibration
Results for the analysis of a silicon nitride bearing using RSF’s obtained from
standard solutions are given in Table 5-1. For comparison, results obtained with EPMA
are provided for the three most concentrated elements (Mg, Co, Al) present in the silicon
nitride. Because of the inferior sensitivity of EPMA, these were the only elements
detectable in the sample by this method. The agreement between the techniques was
excellent, suggesting that the use of solutions for calibration is effective for the analysis
of ceramic materials. The precision of the techniques was similar, with % rsd’s (n=10 for
LA and n=5 EPMA) ranging from 2 to 18 %.
In addition to solution-based calibration for silicon nitride ceramics, the use of
NIST 611 glass as a calibration standard was investigated. In this work, a second bearing
was analyzed. Similar to the procedure used for solutions, RSF’s (analyte/Co) obtained
from the glass sample were used to determine the concentration of analytes in the ceramic
sample. A smaller number of elements was used in this study because of the limited
number of certified elements in the glass. In several cases, the noncertified values
provided by NIST were used (Co, Cu, and Zn). The results are presented in Table 5-2

79
Table 5-1. Comparison of results obtained with LA-ICP-MS using
solution calibration and EPMA.
Analyte
LA-ICP-MSa
Concentration (ppm)
EPMAb
Concentration (ppm)
Mg
7200 +/- 400
7300 +/- 500
Co
1420+/-60
1300 +/- 200
A1
2230 +/- 60
2400 +/- 50
Ni
79 +/- 6
Cu
28 +/- 3
Sr
8+/- 1
Mo
12 +/- 1
Ba
16+/- 1
La
7+/- 1
Nb
38 +/- 6
Y
18+/-2
a Confidence interval (95 %) based on 10 measurements.
b Confidence interval (95 %) based on 5 measurements.

80
Table 5-2. Comparison of results using solution and glass-based calibration.
Analyte
Results from
solution RSF’s
(ppm)ab
Results from
glass RSF’s
(ppm)8’b
Results from
glass sensitivities
(ppm)8
Mn
103 +/- 3
123 +/- 5
120+/- 15
Co
1540+/-60
1380+/-90
1400 +/- 200
Ni
58 +/- 2
65 +/- 3
63 +/- 8
Cu
9+/-2
10+/-2
9+/-2
Zn
9.2 +/- 0.7
13 +/- 1
13 +1-2
Sr
7+/- 1
9+/- 1
8+/-2
a Confidence intervals (95 %) based on 10 measurements.
b Results for Co based on Co/Ni relative sensitivity factor (RSF).

81
and demonstrated that similar results were obtained for both solution- and glass-based
calibration.
The last row of this table represents an extension of the use of glasses for
calibration. In this case, the absolute analyte sensitivites (intensity / concentration) from
laser ablated glass were used with a correction factor to account for differences in the
ablated mass between the glass and ceramic sample. The value of the correction factor
was determined by measuring the 2> differences in nominal silicon concentration of the samples. It was determined that the
ablated mass was four times greater for the glass than the ceramic, under similar
sampling conditions. Not suprisingly, the measured analyte concentrations agreed with
those obtained by the other methods. The precision of these measurements was worse
(12-15 % rsd), since the measurements involved non-normalized sensitivity values
obtained from glass.
It would have been beneficial to measure an isotope of silicon along with the
analytes to directly compensate for variations in ablated mass; however, this was not
feasible with the present system. The use of solutions produced high background levels
(mainly oxides) and precluded the measurement of 29Si and 30Si with the multiplier
detector. This could be remedied, at least in part, by incorporating a desolvation
apparatus to further decrease the solvent loading; however, this was not attempted in this
study. Alternatively, independent measurements of ablated mass, such as the use of
acoustic waves generated by the ablation process (89), or measurement of the scattering
signal produced by the mobilized ablation product (90) could prove useful for the
analysis of ceramic materials with glass-based calibration. The use of independent mass

82
normalization would eliminate the need for internal standardization and provide a more
powerful method for analyzing ceramic materials based on glass standards. Light
scattering for mass normalization with glass, ceramic, and soil samples will be presented
in Chapter 6.
Measurement of Mass Ablated and Estimation of System Efficiency
Profilometry and weight loss measurements were performed on the silicon nitride
samples to determine the amount of material removed by the ablation process and to
determine the efficiency of the LA-ICP-MS system. As previously addressed (Chapter
3), sample translation resulted in a larger mass removal rate and subsequently higher
sensitivities. Profilometry measurements (Figure 5-4) on the craters shown in Figure 3-8
indicated that the volume ablation rate was nearly double in the case of sample translation
compared to single spot sampling (3 pL/s compared to 1.6 pL/s). These corresponded to
mass ablation rates of 9 ng/s and 5 ng/s, respectively. The higher ablation rate was due to
the fact that a partially pristine surface was being ablated by each laser shot with
translation of the sample. The ablated masses agreed reasonably well with those obtained
by weighing the sample before and after approximately 50,000 laser shots. The observed
ablation rates were used to determine a system efficiency for the sintering aid Mg (Mg
ions detected / Mg atoms removed). In these measurements, the 24Mg signal was
continuously monitored during the ablation process. The total integrated signal was then
divided by the estimated mass of Mg ablated. Efficiency of detection for Mg was
estimated to be 1 in 500,000. The efficiency of detection for analytes is a function of the
ablation process, transport process, ionization in the ICP, and mass spectrometer
response.

Vertical Displacement (pm)
Figure 5-4. Profilometry traces of craters in silicon nitride with and without translation of the sample.
00
UJ

84
Conclusions
The use of both solution and glass-based calibration has been demonstrated for
the analysis of silicon nitride bearings. The results obtained from solutions were verified
with a complementary technique, EPMA. Use of glasses for calibration provided similar
results to those obtained from solutions and may prove useful in the future for a wide
variety of ceramic materials. A large number of standard glass materials are presently
available, making them very attractive as potential calibration materials for ceramics.
The ability to independently measure differences in the mass ablated for glass and
ceramics would greatly improve the utility of glass-based calibration.

CHAPTER 6
INVESTIGATION OF LIGHT SCATTERING
FOR MASS NORMALIZATION IN LA-ICP-MS
Introduction
To account for variations in ablated mass and improve the precision and accuracy
of laser ablation measurements, the signal from a matrix element is commonly used as an
internal standard (38, 45, 46, 70). This improves the precision, provided the spatial
distribution of the internal standard and the analyte are similar, and the elements behave
similarly in terms of ablation, transport, and ionization in the ICP. To improve the
accuracy of LA-ICP-MS measurements, it is also required that the concentration of the
internal standard be known. In many cases, this information is not readily available or
can vary significantly, such as with many geological materials and multi-layered
ceramics. In such instances, there is a need for independent measurement of the mass
ablated.
Several techniques that independently measure differences in the mass ablated
have been investigated (89-91). Pang et al. demonstrated that the acoustic wave
generated by the ablation process could be used to normalize analyte signals in LA-ICP-
MS (89). Their work resulted in modest improvements of the precision compared to the
non-normalized signals from steel and aluminum alloy samples. Other work has
involved measurements on the mobilized ablation products. Richner et al. measured the
light loss caused by scattering from ablated material as it passed through a cylindrical
85

86
glass tube (91). They reported that normalization of analyte signals from cast iron
samples based on this technique was comparable to the use of an internal standard.
Similarly, Tanaka et al. used the light scattering signal produced from the transported
aerosol to normalize analyte signals (90). In their work with zirconium alloys, they also
reported an improvement in precision that was comparable to internal standardization.
In this work, the use of light scattering for normalization with several matrices,
including brass, glass, soil, and Macor® (Corning Glass Works, Corning, NY) ceramic
has been investigated. It is demonstrated that a simple light scattering system can be used
for mass correction with a variety of materials. Normalization based on light scattering
results in improved precision of LA-ICP-MS measurements; however, this method of
normalization is not as good as the use of an internal standard. The major strength of the
present technique is that it did not require any knowledge of the sample homogeneity or
concentrations of elements. This makes it potentially useful for the accurate analysis of
inhomogeneous samples as well as those for which standards are not readily available.
This latter aspect has been studied by examining the effectiveness of the light scattering
system for normalization between different types of samples (glass, ceramic, and soil).
Experimental
The LA-ICP-MS instrumentation (Chapter 3) and operating conditions (Chapter
4) were previously described. One difference was the use of a smaller ablation cell. It
consisted of a 3 cm i d. Plexiglas tube with a quartz window for transmission of the UV
laser beam. The total volume of the new ablation cell was ~ 50 cm3. The smaller cell
was used to produce a denser ablation aerosol for larger scattering signals.

87
The scattering cell (Figure 6-1) was placed in-line with the transport tubing. It
was placed approximately 0.5 m from the ablation chamber. The body of the scattering
cell was a 4-way Swagelok®(Crawford Fitting Co., Solon, OH) connector with a 1.5 mm
hole drilled in the top for measurement of the particle scattering at 90°. Smaller angle
viewing would result in higher scattering intensities, but the present arrangement was
chosen for its simplicity. Windows of the scattering cell were mounted at Brewster’s
angle on 3 cm extension arms made of stainless steel tubing. The arrangement produced
less stray-light scatter from the windows and interior of the cell. A 3 mW polarized
HeNe laser (Aerotech, Inc., Pittsburgh, PA) operating at 632.8 nm was used as the
scattering source in these measurements. To improve the quality of the output beam, the
laser was spatially filtered before the scatter cell with an adjustable iris.
An end-on photomultiplier tube (R647, Hamamatsu Photonics K.K., Japan) was
used to collect the scatter produced as the ablation aerosol passed through the cell. The
entire scatter assembly was enclosed in a light-tight box. An alternative arrangement,
which would eliminate the need for darkness, would be to use an interference filter that
would transmit only a very narrow region of light centered at the HeNe wavelength.
Signals from the PMT were amplified by a current to voltage amplifier (Keithley
Instruments, Inc., Cleveland, OH) with a gain of 106 V/A and a rise time of 100 ms. The
amplifier output was fed to a computer interface module (Stanford Research Systems,
Inc., Palo Alto, CA) which was triggered at 100 Hz. Normalization measurements were
made by integrating the scatter signal over a given number of shots, as well as averaging
the “steady-state” level produced during continuous ablation of translated samples for a
total of 200 shots (40 s).

Beam Dump
(b)
From Ablation
Chamber
To ICP-MS
Window
Figure 6-1. Schematic of light scattering set-up ((a) top view and (b) side view).

89
NIST glass (611, 612, 614, and 617) and brass (cl 101, 1102, cl 109, cl 110)
standard reference materials were used to assess the utility of the scattering system for
normalization of analyte signals within a particular matrix. In the glass samples, the 88Sr
isotope was measured with the ICP-MS as well a minor isotope of a major matrix
constituent (43Ca) in order to compare the results from scatter normalization with those
obtained by internal standardization. The concentration of Sr in the glasses ranged from
41.72 ppm to 515.5 ppm. For the brass samples, the 68Zn isotope was measured.
Concentrations of Zn in these standards ranged from 15.2 % to 30.3 %. In the inter¬
matrix studies, NIST 611 glass, NIST 2704 soil, and a Macor® ceramic disc were used.
The soil samples were pressed into pellets, without the addition of a binder at a pressure
of 35 MPa, for laser ablation analyses. In these studies, the 28Si isotope was measured
since silicon is the major matrix element in each of the matrices. The amount of silicon
in the glass, Macor®, and soil was 34 % (nominal), 21.5 % (nominal), and 29.66 %
(certified), respectively.
The nature of the particulate material produced from laser ablation of glass, soil,
and Macor® samples was studied by collecting ablated material on 0.3 p.m membrane
filters (Millipore Corp., Bedford, MA). Because of the significant pressure drop created
with the in-line filters, it was necessary to use a vacuum pump on the back side of the
filter to eliminate flow restrictions. Flow conditions, identical to those used in the
normalization experiments, were achieved by adjusting the pump valve and measuring
the Ar flow rate with a rotameter. After collection, filters were glued onto mounts with
conductive carbon paint and coated with AuPd for examination with a scanning electron
microscope.

90
Results
Scatter Signal Normalization and Comparison with Internal Standardization
Initial experiments with the scattering system were performed to determine if the
scattering signals correlated with signals from the ICP-MS. Figure 6-2 illustrates typical
(a) scatter and (b) mass spectrometric signals obtained from 50 laser shots on NIST 611
glass. It should be mentioned that the actual transit time to the scattering cell was ~1 s,
and ~3 s to the ICP-MS. The large number of spikes present in the scatter signal was
assumed to be due to larger particles passing through the scatter cell based on visual
observation. To evaluate the efficiency of the system for detecting changes in the mass
removed, ablation was performed with different laser energies and focus positions. The
total difference in mass ablated over these conditions ranged over approximately one
order of magnitude. A plot of the average scatter and 28Si mass signals (Figure 6-3)
demonstrated that the two signals were highly correlated (R = 0.98). In this example,
differences in the mass ablated were generated by changing the laser sampling parameters
(i. e. focus and pulse energy); however, these differences occurred naturally over the
series of glass standards under identical sampling conditions. This was probably due to
differences in the surface characteristics of the glasses, since they were all completely
opaque at 266 nm. Similar behavior has been reported by Mermet and co-worker in the
laser ablation of glasses with a 266 nm Nd:YAG (92).
In the absence of normalization, differences in the mass ablated between the
samples would not allow small changes in analyte concentration to be detected. This is
illustrated by a series of calibration curves for Sr in NIST glasses (Figure 6-4), where in
(a), the non-normalized analyte signals are plotted; in (b), the scatter normalized signals

Si Intensity (cps) Scatter Intensity (arb)
91
Figure 6-2. Plots of (a) scatter and (b) mass spectrometric signals
obtained from 50 laser shots on glass surface.

Si Signal (cps)
Figure 6-3. Correlation plot of mass spectrometric and scatter signals from NIST glass.
.•
VO
K>

Sr Signal (counts/s)
93
u
a
Im
5/5
4.5x10
4.0x10
3.5x10
3.0x10
2.5x10
T
30
140
a
U
b
to
0.30
0.25
0.20
0.15
30
T
Concentration (ppm)
Figure 6-4. Calibration plots for Sr in NIST glass samples (a) without
normalization, (b) with scatter normalization, and (c) with
internal standardization.

94
are used; and in (c), the analyte signals are corrected by an internal standard (43Ca).
Without the use of normalization by scattering or internal standardization, a linear
regression R value of 0.77 was obtained. R values obtained with the normalization
techniques were both greater than 0.999. The precision of the scatter normalized
measurements, for 5 replicates and 200 laser pulses/replicate was 5-12 % rsd. This was
not as good as the precision obtained with an internal standard (3-6 % rsd); however, the
use of light scattering is not limited to homogeneous samples, or samples in which a
suitable internal standard is available. This includes samples where the major matrix
constituents do not have a minor isotope, resulting in ion intensities which would saturate
the multiplier detector, or the isotope is interfered with from a sample constituent or some
background species. The precision of the non-normalized signals ranged from 5-19 %
rsd.
Precision of the scatter normalized measurements was slightly worse than that
reported by Tanaka et al. (90), although a direct comparison cannot be made because in
the present work, a low energy UV laser was used for ablation; while in their work, a 150
mJ Nd:YAG operating at 1064 nm was used to ablate zirconium alloys. This should have
resulted in significantly larger quantities being ablated, although the mass of material
removed was not reported.
Although the main focus of the present work was to evaluate the effectiveness of
scatter normalization with nonconducting matrices, NIST brass samples were also
analyzed. Figure 6-5 demonstrates the use of scatter normalization for the calibration of
Zn in these standards. The calibration plot exhibited a higher degree of correlation when
the 66Zn was corrected by the measured scatter signal (R = 0.999) as shown in (b), than

Zn Signal / Scatter Zn Signal (counts/s)
95
3.0x10
2.5x10
2.0x10
1.5x10
1.0x10
7x10
6x10
5x10
4x10
3x10
2x10
8
Zn Concentration (%)
Figure 6-5. Calibration plots for Zn in NIST brass samples (a) without
normalization and (b) with scatter normalization.

96
when the analyte signal alone was plotted (a) (R = 0.94). Precision of the measurements
was 6-11 % rsd (n=5, 200 pulses/measurement) for the normalized signals and 8-20 %
rsd for the non-normalized values. Not only did the use of light scattering improve the
\
precision of the measurements within a given sample, but it also accounted for the
different masses ablated between samples as evidenced by the improved fit of the
regression line.
Comparison of Scatter Normalization for Glass. Soil, and Macor® Ceramic
A comparison of glass, soil, and Macor® was made to assess the utility of the
scattering system for corrections of mass ablated between these matrices. This was
initiated, in part, by the desire to use glass standards for the calibration of ceramic
materials (Chapter 5). In Chapter 5, it was demonstrated that glasses could be used to
analyze silicon nitride ceramic bearing, provided that differences in ablated mass were
accounted for. To study whether light scattering could be used for mass normalization
between matrices, a glass (NIST 611) disc, soil pellet (NIST 2709), and Macor® disc
were ablated to investigate the relationship between mass removed and scattering signal.
Correlation plots, similar to Figure 6-3, are shown in Figure 6-6 for the three
different matrices. In this figure, the MS signals have been normalized by the amount of
silicon present in the samples to allow for a direct comparison of the matrices. Changes
in the mass ablated were achieved by varying the energy of the laser. Slightly higher
laser energies were used for the Macor® sample, since a smaller quantity of material was
ablated for a given energy compared to the glass and soil samples. The laser energy
ranged from 0.1 mJ to 0.7 mJ. Correlation between the scatter and MS signals was good
for the glass (R = 0.94) and Macor® (R = 0.96) samples, but slightly worse for the soils

Si Signal / % Si
Scatter Signal (arb.)
Figure 6-6. Correlation plot of mass spectrometric and scatter signals from
glass, macor, and soil.
-0

98
(R = 0.88). Most importantly, the plot indicated that the scattering signal effectively
accounted for differences in the mass ablated between the glass and ceramic samples as
evidenced by the proximity of the data points for these materials. The similar scatter-MS
signal relationship of glass and Macor® is more clearly illustrated in Figure 6-7, where a
single regression line could be drawn through the set of data points with an R = 0.97.
The scatter-MS signal relationship was significantly different for the soils. In this case, a
larger scattering signal was produced per unit of matrix removed, which resulted in the
soil data points lying closer to the scatter axis.
The nature of the particulate matter produced from ablation of these materials
helped explain the experimental results. Scanning electron micrographs (SEMs) of the
collected particles are shown in Figure 6-8. The pictures clearly demonstrated that the
particles produced from (a) glass and (b) Macor® were both smaller and more similar
than those from the (c) soil samples. The similarity in particle sizes for these matrices
allowed for effective mass correction based on the scattering signal. The larger particle
sizes produced from the soil sample resulted in a significantly different Scatter-MS signal
relationship. This difference was not surprising; however, the larger soil particles, to a
first approximation, should have produced a smaller scatter signal per unit mass
compared to the other matrices. The opposite behavior that was observed in this study
was most likely due to the fact that the larger particles the scatter cell “saw” were not
efficiently transported to the ICP, and would therefore not contribute to the MS signal.
Transport efficiency as a function of particle size was discussed in Chapter 3. This is a
limitation of the present system and could be alleviated by placing the scattering cell

Si Signal / %Si
Scatter Signal (arb.)
Figure 6-7. Correlation plot of mass spectrometric and scatter signals
for glass (squares) and macor (circles).
VO
VO

100
Figure 6-8. Scanning electron micrographs of particles collected
for (a) glass, (b) macor, and (c) soil.

101
closer to the ICP torch, resulting in a more representative view of the material introduced
to the ICP-MS.
Figures 6-9 (a-c) are higher magnification images of particles collected for glass,
ceramic, and soil, respectively. Most of the particles produced from glass and Macor®
were nearly perfect spheres. These were formed by cooling of liquid material and
suggested that the bulk of the ablated material was melted and then explosively removed
from the sample (93). Angular particles were also produced as a result of the mechanical
and thermal shock of ablation, as well as a small amount of amorphous material that was
most likely condensate from the gas phase. Such particle classifications have been
described by other researchers as well (93, 94). The ablation product from soil was
dominated by larger particles that resulted from mechanical and thermal shock, although
each class of particle was seen. The greater presence of large particles was reasonable
since these pressed pellets did not possess the mechanical rigidity of the glass and
Macor® samples.
Weight loss measurements were performed for the glass and Macor® samples to
verify that the mass ablated agreed with the experimental results. The samples were
weighed before and after 40,000 laser shots at a laser pulse energy of 0.7 mJ. The
average mass ablated for glass and Macor® was 16 ng/shot and 6.1 ng/shot, respectively.
This agreed reasonably well with the scattering signals produced for ablation at this laser
energy, and demonstrated that the scattering system was capable of measuring relatively
small quantities of ablated material passing through the scatter cell. Scattering sensitivity
could be improved with a smaller observation angle and/or a shorter wavelength laser;
however, the present system was sufficiently sensitive and easily set-up.

102
Figure 6-9. Higher magnification images of (a) glass, (b) macor and
(c) soil particles. Note difference in scale for soil.
(Bar represents 1 pm in (a) and (b); 10 pm in (c)).

103
A simple measurement was made to assess whether matrix effects, in terms of the
composition of the ablation product, were significant between the glass, Macor®, and soil
samples. In a stricter sense, matrix effects also refer to the differences in ablated mass.
In this instance, only the representativeness of the ablation product was of interest, since
the scattering signal could be used for mass correction with the glass and Macor®
samples. Aluminum, which was a percent level constituent in each sample, was
measured in addition to silicon for each matrix. The RSF’s (Al/Si) obtained for the glass
and Macor® were identical (RSF = 2.30), while soil produced a significantly different
value (RSF = 3.40). This illustrated that matrix effects associated with the composition
of the ablation product were not significant between the glass and Macor® ceramic, and
provided further evidence of the utility of glasses for calibration of ceramic materials.
Conclusions
The use of light scattering for LA-ICP-MS signal normalization was demonstrated
for a variety of matrices. For measurements between matrices, it was important that the
particle sizes generated by ablation were similar, as was demonstrated for glass, Macor®,
and soil samples. The use of this technique was not as effective as the use of an internal
standard in terms of the precision obtained. Light scatter normalization, however,
requires no a priori knowledge of the sample heterogeneity or elemental composition. It
provides a mass normalization alternative when a suitable internal standard is not
available and its potential utility for the analysis of ceramic materials based on glass
standard reference materials has been demonstrated.

CHAPTER 7
ANALYSIS OF SOIL AND SEDIMENT SAMPLES
Introduction
The elemental analysis of soils is important for several reasons, with one of the
most important being the identification of contaminants and establishment of levels of
toxic elements present in the soil. The well-known toxic elements include arsenic,
beryllium, cadmium, mercury, lead, antimony, tellurium, and thallium (7). In addition,
many elements that are essential at low levels (e. g. cobalt, iron, zinc, and copper), become
toxic at high levels (7). Therefore, sensitive and accurate techniques are required for the
characterization of soil samples. An additional complicating factor in trace element
toxicity is the symbiotic nature of coexisting metal species. For example, the presence of
selenium has been shown to provide some protection against heavy metal toxic effects (7).
This emphasizes the importance of multielement techniques for analyzing soil and other
environmental samples.
The refractory nature of geological materials complicates the determination of
elememental constituents in soils (7). Traditional methods of analysis rely on
decomposition of the sample, typically through microwave digestion, and analysis of the
resulting solution by flame or furnace atomic absorption spectrometry (AAS) or ICP-
AES/MS. Direct methods of analysis are advantageous because of the elimination of time-
consuming sample preparation steps and the risk of sample contamination from chemical
104

105
reagents. X-ray fluorescence is commonly used for the analysis of soils; however, the
technique lacks the sensitivity to determine many low level species of interest.
Recently, laser sampling techniques have gained popularity for analyzing soil
samples, since little to no sample preparation is required. Laser induced breakdown
spectroscopy (LIBS) is well suited for field based analysis of soil samples, as evidenced by
the number of recent publications on this application (6, 95, 96). The technique, however,
suffers from more severe matrix-effects and poorer sensitivity when compared to LA-ICP-
MS. Relatively few applications of LA-ICP-MS for analyzing soils have been reported
(97-100). These applications have used either soil standards (97, 98, 100) or fused beads
of soil (99) for analysis.
In this work, soil samples were analyzed by LA-ICP-MS and the utility of solution-
based calibration was investigated. Sampling strategy and choice of the internal standard
were found to be critical for obtaining accurate and precise results. Accuracy was
reasonably good (+/- 20 %) for most elements studied; however in some cases, significant
differences (around a factor of 3) were observed. To account for these discrepancies,
several potential sources of error were investigated. These included speciation effects,
organic content of the soil, and particle size effects. In addition, standard additions was
studied for analyzing trace elements in soils. The effect of particle size on the utility of
standard additions was studied.
Experimental
The LA-ICP-MS instrumentation (Chapter 3) and operating conditions (Chapter 4)
have been previously described. The laser was operated at 5 Hz with typical pulse
energies of 0.7 mJ. Single spot sampling and translation sampling were both used in this

106
work and a comparison of the two will be made. Typical analysis times were 60 s (signals
averaged over 300 laser shots).
The solution calibration methodology was discussed in Chapter 4. Several soil and
sediment standard reference materials were analyzed to determine the applicability of
solutions for calibration of these materials. These materials included NIST 2704 (Buffalo
River Sediment), NIST 2709 (San Joaquin Soil), NIST 2710 (Montana Soil), NIST 2711
(Montana Soil), High Purity Standards (HPS) (Charleston, SC) Sandy Soil B, and High
Purity Standards Loam A. The samples were pressed into pellets at a pressure of 35 MPa
without any binder for analysis. A single multielement solution containing 2 ppm of Ca
and 20 ppb of all other analytes was used to determine solution relative sensitivity factors
(RSF’s). RSF’s were used, as discussed in Chapters 4 and 5, to determine analyte
concentrations in the soils. Choice of an appropriate internal standard was found to be
critical in the analysis of soil samples. This will be described more thoroughly in the
experimental results.
To study the effect of organic content of the sample on LA-ICP-MS results,
cellulose binder (Spex CertiPrep Metuchen, NJ) was added to NIST 2709 soil in varying
proportions (10 % and 20 % w/w). The 1 g soil and binder samples were mixed in plastic
vials with a Spex Mixer/Mill (Model 8000) for 30 minutes and then pressed into a pellet.
Additionally, an inorganic sample (NIST 1633 Coal Fly Ash) was studied. With the coal
fly ash standard, a sample mixed with cellulose binder (10 % w/w) was prepared, as well
as one mixed with high-purity graphite (20 % w/w). The 1 g samples were pressed into
pellets at 35 MPa.

107
Because elements exist in a variety of forms (e. g. carbonates, silicates, oxides,
etc.) in soils, the effect of speciation on LA-ICP-MS results was studied. Samples
containing various compounds of Ba were prepared in a sand matrix (Mallickrodt Baker,
Inc. Paris, KY). Compounds of Ba (nitrate, oxide, chloride, carbonate, and sulfate)
(Aldrich Chemical Company, Inc. Milwaukee, WI) were added to 10 g of sand to produce
samples with a total Ba concentration of -750 ppm. In addition, Ni (as sulfate) was added
at the same concentration (-500 ppm) in each sample for use as an internal standard.
Each of the mixtures was ground and homogenized in an alumina grinding vial (Spex
Model 8003) for 30 minutes. A 0.9 g portion of the sample was then mixed with 0.1 g of
cellulose binder for 30 minutes before being pressed into a pellet. To study the matrix
dependence of speciation effects, Ba (and Ni as internal standard) samples were prepared
in a graphite matrix. Direct grinding and mixing of the Ba and Ni compounds in graphite,
however, did not produce homogeneous samples. It was necessary to prepare
concentrated mixtures (10 % w/w) of Ba and Ni in sand, and then dilute this mixture in
graphite to give final concentrations of-750 ppm for both analytes.
To determine whether differences (composition, measurement precision, etc.) were
observed for different particle size fractions, NIST 2704 and HPS Sandy Soil B samples
were sieved into two different size fractions (<35 jam and 35 to 60 pm). One gram
samples of each fraction were then pressed into pellets for analysis.
The utility of standard additions for accurate analysis of trace elements in soil was
briefly assessed. Multiple standard additions of Co to NIST 2709, NIST 2710, and NIST
2711 were made by spiking the soil samples with Co solutions. The concentrations of the
solutions were adjusted so that the same volume (3 mL) was added to 1 g soil samples.

108
The mixture was then dried at 110 °C for several hours in glass vials. Dried soil samples
were then transferred to mixing vials and shaken for 10 minutes to ensure homogeneity.
They were then pressed into pellets. To study the effect of particle size on standard
addition measurements, sand samples of various particle sizes were spiked with a
multielement solution. To generate a range of particle sizes, sand samples (~10 g) were
ground for 1 to 2 minutes. The ground material was then transferred to the sieve and four
particle size fractions (<35 pm, 35-60 pm, 60-80 pm, and >80 pm) were obtained. One
gram samples of the different particle size fractions of sand were placed in glass vials,
spiked with solutions containing Co, Ag, Y, Rb, W, Ba, and Pb, and dried at 110 °C for
several hours. The sand samples were then mixed with 10 % (w/w) cellulose binder for 30
minutes and pressed into pellets for analysis.
Results
Spot Sampling Versus Line Sampling
Single spot and translational line sampling were studied for the analysis of soils.
Signals were typically a factor of 4 or 5 larger when the sample was translated. This
increase was greater than what was observed with glass and ceramic samples, most likely
because a deep crater was formed more quickly in these compacted particulate samples.
A comparison of the results obtained with both sampling strategies is shown in Figure 7-1
for NIST 2704. These figures represent the analyte responses for Ba and Ni relative to Ca
from a total of 25 scans obtained from 25 different spots on the sample, or at different
lines (5 scans/line) produced from translation of the sample. Cu and Co were also studied
and produced similar results. The relative analyte responses from both sampling methods
were similar in most cases; however, significant deviations from the mean (as much as a

109
ns
H
22
â– Si
C
B
"O
u
N
o
Z
Scan Number
Figure 7-1. Comparison of relative analyte signals for (a) Ba and (b) Ni
using both single spot and translational sampling.

110
factor of 5) were observed in some scans when sampling at a fixed location. This must
have resulted from local inhomogeneity in the soil samples, and clearly indicated the need
to sample a large enough portion of the soil to ensure accurate and precise measurements.
Measurement of these local inhomogeneities is required for accurate measurement of the
bulk composition; however, their presence can significantly affect the precision of
analyses. From the results in Figure 7-1, it appeared that translation of the sample
effectively averaged out the inhomogeneities during a typical 1 minute analysis, which
should result in more precise measurements. The absence of large deviations in the
relative analyte signal (analyte/Ca) was almost certainly due to the larger mass ablated
with sample translation. The average precision (% rsd) of the relative intensities for the
four analytes studied was 8.8 % with translation sampling and 37 % with single spot
sampling. Since higher sensitivity and better precision were obtained with translation of
the sample, this mode of sampling was used for all subsequent measurements.
Solution-Based Calibration
Solution-based calibration required selection of an appropriate internal standard
with a known concentration in the sample. For the analysis of glass samples (Chapter 4),
it was demonstrated that reasonably accurate results could be obtained whether a trace
element (e. g. Sr) or major matrix constituent (e. g. Ca) was used as the internal standard.
Because of greater inhomogeneity in the soil samples, measurement of a minor isotope of
a major matrix consitituent (Ca) was used to provide an acceptable level of precision.
Using an isotope of Ca (43Ca or 44Ca), which was present at levels of 1-3 % in the soils
studied, as the internal standard resulted in typical precision values (% rsd, n = 10) of less
than 10 %. Precision values were not consistent among the soils analyzed. Measurements

Ill
on the NIST 2704 sample exhibited the poorest level of precision, with % rsd values
consistently around 10-12 %. With the other soils, the precision was typically around 6-7
% rsd. Measurement precision is obviously dependent on the concentration of analyte in
the sample and the amount of material ablated; however, this cannot explain the poorer
precision observed with the NIST 2704 sample. This sample contained levels of trace
elements and Ca that were similar to the other samples. In addition, a similar mass (~ 50
ng/shot) was ablated for all of the soils studied. The differences in precision that were
observed might have been related to differences in the particle size distribution for the
NIST 2704 sample compared to the other soils. The effects of particle size on
measurement precision will be examined later in this chapter.
Selection of the internal standard was found to be an important factor in obtaining
accurate results for soils. Initially, Ca was used as the internal standard because it was
present at significant levels in all of the samples and possessed minor isotopes that could
be measured with the multiplier detector. Calcium was used for the glass calibration
studies and was found to provide accurate results for all elements studied. This was not
the case with the soil samples. A comparison of the RSF’s for several elements (V, Co,
Ni, Cu, Zn, Sr, Ag, Ba, and Pb) in NIST 2709 and NIST 2704 soils using 43Ca as the
internal standard is given in Figure 7-2. These plots have been normalized to solution RSF
values; therefore, the accuracy of the solution calibration method can be directly assessed
by comparing the normalized soil RSF’s to a value of one (a value of one signifies that
solution and soil determined RSF’s are identical). For NIST 2709, the measured
concentrations of V, Co, Ni, Cu, and Zn were between 40 and 60 % higher than the
certified values for these elements; Ba and Sr were about 60 % lower than the certified

Figure 7-2. Comparison of RSF’s for NIST 2709 and NIST 2704 soils.
Plots have been normalized to solution RSF values.

113
values; and Pb was within 10 % of the certified value. For NIST 2704, different results
were obtained. There was a systematic increase in the measured concentration for all of
the analytes studied in this sample. Systematic changes were observed for the other soils
as well. Because errors in the measured concentrations were largely systematic, rather
than random, it is believed that they were the result of a spectroscopic interference by
aluminum oxide at m/z 43. Based on these observations, 43Ca was deemed a poor choice
for internal standardization since the signal at this mass was affected by the levels of
aluminum in the sample. The use of 44Ca was also investigated for internal
standardization. Similar systematic differences in the measurement accuracy were
observed with this isotope as well, most likely as a result of interferences due to silicon
oxide or carbon dioxide at m/z 44. Using an isotope of silicon (28Si, 29Si, or 30Si) as the
internal standard worked well for the analysis of minor elements in soil (Figure 7-3);
however, none of the silicon isotopes could be measured on the multiplier detector due to
saturation of the detector. Use of the multiplier was required for trace element analyses.
Because of the interferences on Ca istopes and the lack of any other suitable internal
standard that could be used over the whole series of soils, a minor sample constituent was
used for further solution calibration work.
The use of trace elements as internal standards produced more accurate
measurements for the soil samples. This is illustrated in Table 7-1, which compares the
measured concentrations for several analytes in NIST 2704, NIST 2709, and NIST 2711
soils using the solution calibration method with both 43Ca and 60Ni as internal standards.
The RSF values using Ni as the internal standard were determined by dividing the
analyte/Ca RSF by the Ni/Ca RSF. Significantly more accurate results were obtained

Normalized Ba Intensity Normalized Mg Intensity
114
Figure 7-3. Calibration plots for (a) Mg and (b) Ba in soil samples
using 28Si as an internal standard.

115
Table 7-1. Measured concentrations in soils using solution calibration
with 43Ca and 60Ni as internal standards.
NIST 2704
NIST 2709
NIST 2711
Cert.
Meas."
Meas. “
Cert.
Meas."
Meas."
Cert.
Meas."
Meas."
43Ca
6°n¡
43 C a
60Ni
43Ca
6°n¡
(ppm)
(ppm)
(ppm)
(ppm)
(ppm)
(ppm)
(ppm)
(ppm)
(ppm)
V
95
210
87
112
160
116
81.6
106
57.8
± 36
±15
± 12
± 10
±9
±4.7
Cr
135
326
135
130
146
106
(47)
59.8
32.5
±48
±22
± 12
±11
±6.7
±4.8
Co
14.0
33.6
13.9
13.4
21.6
15.6
(10)
18.4
9.2
±5.2
±2.6
± 1.3
± 1.4
±2.3
± 1.0
Ni
44.1
int.
int.
88
int.
int.
20.6
int.
int.
std.
std.
std.
std.
std.
std.
Cu
98.6
249
103
34.6
52.3
37.9
114
182
110
±31
± 16
±3.9
±3.8
± 18
± 16
Zn
438
940
390
106
150
109
350.4
471
255
± 140
±72
±9.3
±9
± 56
±38
Sr
(130)
119
49.0
231
107
77.5
245.3
139
75.6
±12
±8.2
±5.3
±5.8
±9
±5.6
Ag
0.41
0.39
.28
4.63
7.2
3.9
±.10
±.08
± 1.3
±0.8
Ba
414
351
146
968
369
268
726
439
240
±48
±34
±27
±23
±40
±30
Pb
161
570
235
18.9
21.1
15.3
1162
3150
1570
± 100
±58
4.2
±3.2
±210
± 140
n = 5, 95 % confidence level

116
using Ni as the internal standard, thus indicating that solution calibration was useful even
for a complex matrix like soil. A limitation of the technique, however, is that the
concentration of some element that can serve as the internal standard must be known in
the sample. This can be addressed by using a complementary technique to determine the
concentration of an element in the sample, or by spiking in a known amount of some
element that is not in the sample at an appreciable concentration relative to the amount
added. A suitable element for spiking could be chosen by doing a survey scan over the
sample of interest.
For the analytes studied, consistent patterns concerning the accuracy of solution
calibration were observed. Most analytes (V, Cr, Mn, Co, Ni, Cu, and Zn) could be
determined with reasonable accuracy (typically +/- 20 %) using a single solution for
calibration; however, the measured concentrations for several analytes (Rb, Sr, Ba, and Y)
were consistently lower (by around a factor of 2 or 3) than their certified values in the
soils. More accurate measurements of the latter elements could be obtained if one element
in this group was used as the internal standard for the other elements. The lower results
obtained for Rb, Sr, Ba, and Y may be due to matrix effects resulting from high levels of
efficiently ionized elements (EIE). EIE’s generally cause a decrease in ICP-MS intensities,
and are most severe for elements with low ionization energies (101). This might account
for the lower results obtained for Rb, Sr, Ba, and Y, since these analytes possessed the
lowest ionization energies of the elements studied. In general, matrix effects in ICP-MS
are difficult to measure and quantify (23). They can often be minimized for particular
analytes through optimization of the ICP-MS operating conditions or selection of
appropriate internal standards (102); however, for multielement determinations, this is

117
often not feasible since all elements do not behave identically in the ICP-MS. Therefore,
improvements with respect to one analyte may adversely affect the measurement of
another analyte. The effect of ICP RF power on RSF’s from solution and soil-based
analytes was briefly studied to see if suitable conditions for analyzing Co, Cu, Sr, and Ba
could be achieved. The results are presented in Figure 7-4. In these plots, Ni was used as
the internal standard and all of the RSF values were normalized to the solution-based RSF
value at 1200 W of RF power. Changing the RF power affected the RSF values slightly;
however, the changes were similar for both the solution and the laser ablated NIST 2709
soil sample. Therefore, no improvement in the accuracy was observed as the RF power
was changed over the range of 1050 to 1350 W.
Although accurate measurements were not obtained in all cases, solution
calibration provided a relatively simple means of estimating analyte concentrations in the
sample. In all cases, the measured analyte concentrations were within a factor of three of
their certified concentrations, and in most cases, much better accuracy was achieved. To
investigate potential sources of inaccuracy in soil samples, element speciation, sample
organic content, and particle sizes effects were studied.
Speciation Effects
Since metals exist as a variety of compounds in soil samples, the effects of speciation
on LA-ICP-MS measurements was examined. In a study involving the detection of Ba
and Pb in soils by LIBS, it was determined that the chemical form (oxide, carbonate,
sulfate, chloride, nitrate) of the analyte affected the sensitivity of LIBS measurements (6).
To see if this might help explain the errors observed for Ba, Sr, Rb, and Y in this LA-ICP-
MS work, the effects of speciation on Ba in a sand matrix was examined. The results for

Normalized RSF(Co/Ni)
118
Figure 7-4. Effect of RE power on RSF’s (solution and soil)
for Co, Cu, Sr, Ba.

Normalized RSF(Ba/Ni) Normalized RSF(Sr/Ni)
119
2.0-T
1.5-
-■—solution
A soil
1.0-
I I £ I I I
I
I 1 I 1 I 1 I 1 I 1 I 1 I r
1050 1100 1150 1200 1250 1300 1350
RF Power (W)
Figure 7-4 continued.

120
Table 7-2. Ba speciation results (sand matrix).
LA-ICP-MS
RSF(Ba/Ni)a,b
LA-ICP-MS
RSF(Bat„tai/Ni)a’c
ICP-MS
RSF(Ba/Ni)M
Oxide
0.89 ±0.09
1.18 ± 0.10
1.66 ±0.02
Nitrate
1.22 ± 0.16
1.64 ± 0.17
1.70 ±0.02
Carbonate
1.14 ± 0.07
1.50 ±0.09
1.68 ±0.04
Chloride
1.33 ±0.09
1.75 ±0.09
1.56 ±0.01
Sulfate
1.19 ± 0.05
1.58 ±0.05
a n = 5, 95 % confidence level.
b laser ablation results using Ba+ signal only.
c laser ablation results using Ba+, Ba2+, and BaO+ signals.
d solution results using Ba+ signal only.

121
these studies are given in Table 7-2. The first column of data compares the RSF(Ba/Ni)
values obtained from the five sand samples. The BaO sample produced a lower value
(statistically significant at 95 % confidence level); however, the decrease was minor in
comparison to the factor of 2-3 lower concentration measured with solution calibration.
To see if the observed decrease could be attributed to the formation of Ba2+ or BaO\ the
signals for these isotopes were included in determining RSF’s (column 2 of Table 7-2).
The same trend was observed; BaO produced slightly lower RSF’s compared to the other
chemical compounds. The RSF values given in the last column of Table 7-2 were
measured from the sand pellets by solution nebulization to ensure that the samples
contained the same levels of Ba and Ni. A 5 mg portion of each sand sample was diluted
in 20 mL of 5 % FINO3 and then heated to dissolve the Ba and Ni species. The solutions
were analyzed and the results indicated that the pellets contained reproducible levels of Ba
and Ni. The sulfate pellet was not included because BaS04was insoluble in the dilute
HNO3 solution.
The effects of speciation were examined in a graphite matrix as well. Only the
oxide, sulfate, and carbonate forms of Ba were studied, with Ni (sulfate) as the internal
standard. No significant differences between the Ba/Ni RSF’s were observed. The
RSF(Ba/Ni) values were 1.45 ± 0.19 for the oxide, 1.48 ± 0.25 for the sulfate, and 1.54 ±
0.27 for the carbonate. Based on the speciation studies with Ba, it was determined that
the form of the analyte did not significantly affect LA-ICP-MS measurements.
Effect of Organic Content
To determine whether organic content of the sample influenced LA-ICP-MS
measurements, both soil and coal fly ash samples were studied. Organic content was

122
studied not only to determine if this might help explain the observed discrepancies for Ba,
Sr, Rb, and Y, but also to determine if the addition of cellulose binder, used in pellet
preparation, affected the accuracy of results. Previous studies using solution nebulization
ICP-MS have indicated that the addition of small amounts of organic solvents can either
increase or decrease analyte signals. Main, et al. found that the addition of glycerol or
methane significantly enhanced signals for some analytes (As, Au, Se, Te, and As), while
other analytes were essentially unaffected (103). These workers attributed the signal
enhancements with addition of carbon to a modification of ionization equilibrium over a
limited energy range (9-11 eV). The ionization energy of carbon (11.2 eV) is slightly
above this range. Longerich reported ICP-MS signal suppression for several elements
when various organic solvents were added to a 1 % HN03 solution. (104). The loss in
sensitivity could be recovered, however, when the nebulizer gas flow rate was reduced
from the level corresponding to the maximum for a 1 % HNO? solution.
Results for the analysis of NIST 2709 soil, which contained 1.2 % carbon (organic
and inorganic), with no binder added, 10 % cellulose binder added, and 20 % cellulose
binder added were compared (Table 7-3). No significant changes in the measured analyte
concentrations were observed with the addition of cellulose binder at levels up to 20 %.
Similarly, a coal fly ash sample was studied to determine if different results were obtained
when 10 % cellulose or 20 % graphite was used to prepare the sample pellet. The results
for these analyses (Table 7-4) indicated that the addition of an inorganic (graphite) or
organic (cellulose) binder produced similar results. The accuracy of these measurements,
based on solution calibration with Ni (98 ppm in sample) as the internal standard, was
around 10-20 % for all analytes studied.

123
Table 7-3. Comparison of RSF’s from NIST 2709 soil with and without binder.
RSFa
No cellulose
RSFa
10 % cellulose
RSFa
20 % cellulose
V
32.8 ±2.7
34.3 ± 1.0
36.2 ± 1.4
Cr
3.24 + 0.30
3.45 ±0.38
3.55 ±0.26
Mn
35.6 + 4.1
35.9 ± 1.9
36.3 ±2.3
Co
36.5 ±2.1
37.4 ± 1.0
40.4 ±3.6
Ni
7.04 ±.36
7.01 ±0.25
7.69 ±0.50
Cu
16.4 ±0.9
16.7 ±0.8
18.6 ± 1.4
Zn
3.48 ± 0.17
3.51 ±0.25
3.73 ±0.33
Rb
11.6 ±0.9
12.1 ±0.69
14.0 ±0.6
Y
16.7 ±2.7
15.3 ± 1.4
16.6 ±3.0
Ba
1.41 ±0.16
1.40 ±0.11
1.39 ±0.08
a n = 5, 95 % confidence level. 43Ca as internal standard.

124
Table 7-4. Measured concentrations from NIST 1633 (coal fly ash)
with different binders using solution calibration.
Cert. Cone, (ppm)
Meas. Cone. (ppm)a
20 % graphite
Meas. Cone. (ppm)a
10 % cellulose
V
214 + 8
239+ 12
244 + 9
Cr
131+2
153 ± 14
160 + 8
Mn
493 +7
571+21
579 ± 43
Cu
128 + 5
143+3
153 ± 10
Zn
210 + 20
223 +9
235 + 15
Cd
1.45 + 0.06
1.65 + 0.27
1.83+0.26
an = 5, 95 % confidence level. 60Ni used as the internal standard.

125
Analysis of Particle Size Fractions
Two different particle size fractions of NIST 2704 and HPS Sandy Soil B were
analyzed to determine whether analytes were distributed similarly in both fractions, and to
determine how particle size affected the precision of LA-ICP-MS measurements when
analyzing particulate samples. The results for NIST 2704 (Figure 7-5) indicated that the
concentrations of the eight trace elements studied were identical, within the experimental
uncertainty, in both particle size fractions. In these plots, the analyte response was
normalized by 43Ca to account for differences in ablated mass. The ablated mass was
slightly higher (-20 %) for the pellet consisting of particles less than 35 pm, compared to
the 35-60 pm pellet. Also, the precision of the measurements was significantly better for
the smaller particle sample (average of 7. 3 % rsd) than the larger particle sample (average
of 14 % rsd). The unsieved sample was included for comparison in Figure 7-5. The
precision of measurements on this sample was intermediate of the two fractions, with an
average % rsd of 10 %. The original material consisted of -75-80 % of particles less than
35 pm.
The HPS Sandy Soil B sample produced different results in terms of particle
composition. The Ca-normalized results (Figure 7-6) indicated that the smaller particles
(<35 pm) had approximately three times more Pb and Ag than the larger particles (35-60
pm). All of the other elements studied contained identical levels of trace elements. These
results are not easily explained, but might be related to the preparation of this standard
reference material. In this soil, the concentrations of most trace elements have been
enriched 10 to 100 times by spraying the sample with an aerosol mist. The samples were
then dried, ground, sieved, and blended. If for instance, the elements were added

Relative Intensity
126
Figure 7-5. Comparison of relative analyte responses from different particle
size fractions. Unsieved material included for comparison.

Relative Intensity
127
Zn
Sr
<35 35-60
Ba
0.020
0.010 _
<35
35-60
Figure 7-5 continued.
Orig.
I
Orig.

Relative Intensity
128
v Co
Figure 7-6.
Comparison of relative analyte responses from different particle
size fractions. Unsieved material included for comparison.

Relative Intensity
129
Zn Sr
Figure 7-6 continued.

130
sequentially rather than simultaneously, a preferential adsorption of Pb and Ag on the
smaller particles might have occurred. The behavior of these samples, in terms of
measurement precision, was similar to that observed with the NIST 2704 sample.
Measurements on the smaller particle sample were characterized by higher precision
(average % rsd of 5.5 %) than the larger particle sample (average % rsd of 12 %).
Precision of the measurements for the unsieved sample was 7.1 %, on average. Based on
these studies, the precision of measuements is significantly affected by sample particle size.
Samples consisting of larger particles produced poorer precision and lower ablation yields
than samples consisting of smaller particles.
Analysis of Soils Using Standard Additions
Standard additions was briefly studied for analyzing Co in three soil samples
(NIST 2709, NIST 2710, and NIST 2711). Although sample preparation time was
increased with standard additions, standard additions possessed several advantages. Strict
matrix-matching was achieved and the concentration of the internal standard was not
required. This latter aspect is a limitation associated with the solution-calibration
technique. With standard additions, an internal standard was still used to correct for
variations in ablated mass. This was necessary to produce sufficiently precise
measurements. In general, standard additions is well-suited for the analysis of soil and
other particulate samples, since an element can be homogeneously spiked into the sample
of interest. Results obtained for the analysis of Co in soils using multiple standard
additions are given in Table 7-5. Only one of these materials was certified for Co (NIST
2709). The measured concentration of Co in this sample was not statistically different (95

131
Table 7-5.
Results from the determination of Co in soils
using multiple standard additions.
NIST 2709
NIST 2710
NIST 2710
Certified
Cone, (ppm)
13.4±0.7
(10)a
(io)a
Measured
Cone, (ppm)
11.711.6
8.510.8
7.510.4
a non-certified value.

132
% level) than the certified value. Measured concentrations for the other samples were
similar to the non-certified values.
In these measurements, the Co signal was normalized by the signal from a matrix
element (Ca). To investigate whether particle size influenced relative signals for analytes
deposited on the surface of particles to an analyte in the particle, several spiked sand
samples were analyzed. The same amount of each analyte was added to four different size
fractions of sand (<35 pm, 35-60 pm, 60-80 pm, and >80 pm). All analyte signals were
normalized to 44Ca, which was present in the sand matrix. The results (Figure 7-6)
indicated that the relative signals for all analytes increased significantly for particles greater
than 35 pm. This was due to the inefficiency of the ablation process at the laser energies
used (~0.5 mJ) in completely vaporizing the larger particles of sand. The deposited
analytes were concentrated on the surface of the particles; therefore, they were
preferentially vaporized relative to the bulk of the particle and produced higher relative
intensities. Onset of this process should be shifted to larger particle diameters as the laser
energy is increased; however, the energy used (0.5 mJ) was close to the maximum output
of the laser system. In addition, particle sizes at which preferential vaporization became
significant would depend on the type of sample being analyzed. For more refractory
materials like ceramics, differences in relative analyte signals would be expected to occur
at even smaller particle sizes. Based on the observed differences in relative analyte signals
for the spiked sand samples, it was clear that particle sizes in the sample could significantly
affect the accuracy of LA-ICP-MS results using the standard additions method. If the
particles were too large to be efficiently ablated, only the surface concentration would be
measured. For standard additions measurements, this would increase the slope of the

Relative Intensity (X/Ca)
Figure 7-7. Effect of particle size on relative analyte signals for sand.
Lk>

134
standard additions plot, and measured concentrations would be lower than what was
actually present in the sample.
Detection limits
Detection limits (3g) for elements in soil were typically in the tens of ppb range
(Table 7-6), provided no major background interferences were present. This was
demonstrated by the relatively poor detection limit for Cr (400 ppb), which resulted from
interference by ArC+.
Conclusions
Reasonably accurate (+/- 20 %) results for most trace elements in soil samples
could be obtained using a single solution standard. Poorer results were obtained for Rb,
Sr, Y, and Ba, most likely because of matrix effects associated with large amounts of
easily ionizable elements. Selection of an appropriate internal standard (e. g. Ni) was
found to be important due to interferences on calcium isotopes. Elemental speciation and
sample organic content did not significantly affect these measurements. Particle size was
shown to play a significant role in terms of the precision of measurements. Measurements
performed on samples that were composed of smaller particles were more precise.
Standard additions was demonstrated to be useful for the analysis of Co in soil samples;
however, studies indicated that particle sizes of the matrix could significantly affect the
quality of these results, both in terms of accuracy and precision.

Table 7-6. Detection limits (3a) for
selected elements in soil.
Analyte
Detection Limit
(PPb)
V
6
Cr
400
Co
3
Ni
30
Cu
20
Zn
200
Sr
8
Ag
4
Cd
30
Ba
30
Pb
200

CHAPTER 8
EVALUATION OF A COMPACT LASER SOURCE
Introduction
The relative merits and applicability of laser sampling for ICP-MS have been
previously discussed. In spite of the numerous advantages of the technique, the cost of
commercially available laser ablation systems can be prohibitive. The most commonly
used type of laser for LA-ICP-MS is the Nd:YAG operating at its fundamental
wavelength (1064 nm) (19, 43, 71, 75-78) or one of its harmonics (532 nm (77-78), 355
nm (77), and 266 nm (66, 69, 70, 75-78)). Although it has been proposed that UV lasers
may be more advantageous for LA-ICP-MS (76-78), the use of doubling crystals
increases the complexity and expense of the system.
The purpose of this work is to report on the use of a compact and inexpensive
Nd:YAG laser for LA-ICP-MS analysis of solids. These lasers, which were developed
primarily for the ophthalmic market, offer a simple, low cost solid sampling option for
ICP-MS users. The device, which measures approximately four inches in length, is
permanently aligned and thus requires no manual adjustments (105). Recently, their use
in a portable Laser Induced Breakdown Spectrometer was reported (95). In this work, the
capability of these compact lasers for LA-ICP-MS analysis of NIST aluminum, copper,
and glass samples is demonstrated.
136

137
Experimental
The laser used in this study was a Nd:YAG (Kigre, Inc. Hilton Head, SC MK-
367) operating at 1064 nm. This laser provided nominal pulse energies of 20 mJ, pulse
widths of ~4 ns, and a beam diameter of around 3 mm. It could be powered from either a
12 V dc power source or a transformer for 110/120 V operation and had an operational
lifetime in excess of 300,000 shots (105). It could be operated in either a manual firing
mode or repetitively fired at a maximum repetition rate of 1 Hz. In our work, the laser
was operated at 1 Hz, which required the use of a water-cooled mount (Kigre
MK235236). The laser was focused onto the surface of the sample with a biconvex lens
(6 cm focal length). Samples were held in a glass ablation chamber which was translated
at ~50 ttm/s on a modified syringe pump, to provide a partially fresh surface for each
laser shot in order to achieve more representative sampling. The total cost of the laser
system and sample stage was under $7000. Ablated material was transported to the ICP
through a 1.5 m length of plastic tubing.
For comparison studies, the Finningan MAT System 266 laser ablation accessory
(described in Chapter 3) was used. It provided a 266 nm Nd:YAG laser beam with a
pulse energy of 0.5 mJ and a pulse width of 10 ns.
The ICP-MS instrumentation has been previously described (Chapter 3)
Operating conditions were similar to those given in Table 4-1.
The samples used in this study included NIST 601-604 aluminum standard
reference materials, in which titanium was present at levels of 120 to 1000 ppm. In the
analysis of these samples, 48Ti was measured as well as 57Fe for internal standardization.
In addition, NIST copper standard reference materials were analyzed for nickel (60Ni

138
measured). The concentration of nickel in the samples ranged from 4.2 to 11.7 ppm. No
internal standard was measured with the copper standards, since the only major matrix
constituent (Cu) produced signals that were too large to measure with the multiplier
detector. To minimize differences in ablated mass between these samples as a result of
different laser focusing conditions, the samples were cut and polished to approximately
the same height.
To demonstrate the utility of this laser for nonconducting samples, NIST glass
samples were also analyzed (611, 612, and 617). In the glasses, 8gSr and 41Ca were
measured, where Sr was present at levels from 41.5 ppm to 515.5 ppm and Ca was a
major constituent of the glass matrix. All of these glasses were used as received with no
surface modification or other preparation.
Results
Scanning electron microscopy was used to examine the laser ablated NIST
aluminum samples as shown in Figure 8-1. Figure 8-la shows the surface of the sample
after 50 laser shots. It can be seen that the largest amount of ablation occurs in a region
approximately 150 pm in diameter, but that the melting of the sample extends well
beyond this area as evidenced by the wavelike structures on the sample surface. Figure
8-lb is representative of typical analysis conditions, i.e., the sample was translated at 50
pm/s while the laser was fired at 1 Hz. The ablation track in this case is around 500 pm
wide and very shallow. Once again, the wavelike structure can be clearly seen and is
indicative of the fact that much of the material is removed by vaporization of the surface
via the plasma that forms on the surface. For comparison, an ablation track formed by
ablation with a 266 nm Nd:YAG is shown in 8-lc. It is evident from these figures that

139
Figure 8-1 Scanning electron micrographs of aluminum surface after (a) 50 shots
from the compact IR laser, (b) while translating the sample, and (c) with
the UV laser while translating the sample. Magnifications of (b) and (c)
are 100X and 240X, respectively. Bar represents 100 pm.

140
the ablation process is much more explosive with the UV laser than with the IR laser.
This is due to the fact that the laser interacts with the sample surface throughout the laser
pulse for UV lasers, and the role of the surface plasma in ablation is minimal. Several
researchers have compared the use of IR and UV lasers for ablation (76-78). Only a
limited comparison can be made in this work, however, since the pulse energy, pulse
width, beam quality, and numerous other factors play a role in the ablation process.
A temporal profile of the Ti signal from NIST 604 is shown in Figure 8-2. This
figure indicates that under the experimental conditions, even the low repetition rate of 1
Hz produced a “steady state” signal. The profile contains a large number of spikes,
which can be attributed to larger particles being vaporized in the ICP.
A linear calibration plot (R = 0.99) was obtained for Ti in the aluminum samples
(Figure 8-3). It was necessary with these samples to measure 57Fe in addition to 48Ti, in
order to account for differences in the mass ablated between samples and within a series
of analyses on one sample. Although it is not exactly an internal standard since its
concentration varied between the samples, we accounted for these differences by
normalizing the signal and thus mimicked the use of an internal standard. It is usually
preferable to use a minor isotope of a major matrix constituent as the internal standard,
but this was not possible because A1 is monoisotopic and produced signal levels that
saturated the detector. The precision of the normalized measurements (n = 5) ranged
from 10 to 13 % rsd, compared to 17 to 24 % without normalization. These values could
be improved by using a shorter dwell time and a larger number of passes, but a minimum
dwell time of 64 ms was required with the Faraday detector. This means that a relatively

Ti Signal (counts/s)
1.2x10 7
1.0x10 7
8.0x10 6
0 20 40 60 80 100 120
Time (s)
Figure 8-2. Temporal profile of Ti signal from NIST 604 aluminum.

Normalized Intensity
1.4
1 1 1 1 1 1 1 ' 1 1 1 1 1
0 200 400 600 800 1000 1200
Concentration (ppm)
Figure 8-3. Calibration plot for Ti in aluminum samples.
K>

143
long period of time was spent at each mass. The detection limit (3a) for Ti in the
aluminum samples was 0.3 ppm.
A linear calibration plot (R = 0.98) was also obtained for Ni in the NIST copper
standard reference materials (Figure 8-4). Precision of these measurements, however,
was relatively poor since no internal standard was used. The % rsd (n = 5) ranged from
12 to 39 %. The detection limit (3a) for Ni was 0.02 ppm.
To assess the use of the laser for nonconducting samples, NIST glasses were
analyzed. A linear calibration plot (R = 0.999) for Sr was obtained (Figure 8-5). As with
the aluminum standards, an additional isotope (43Ca) was measured to account for
differences in the ablated mass. Differences in the ablated mass were as great as a factor
of four between the glasses. This most likely resulted from differences in the 1064 nm
absorption by the glass samples. Variations in the ablated mass between samples was
less than a factor of two with the 266 nm laser, since essentially all the light was absorbed
at this wavelength. The precision of the measurements (n = 5) ranged from 2 to 10 % rsd.
The improved precision of these measurements, compared to the Ti in aluminum
measurements, was at least partially due to the shorter dwell times that could be used
with the multiplier detector. The detection limit (3a) for Sr in the NIST glasses was 0.1
ppm. For comparison, the detection limit with UV laser ablation was 0.05 ppm. The
improvement can be attributed to the larger mass of material that was ablated with the
UV laser, due to increased absorption by the glass. Due to differences in laser ablation
systems and ICP-MS instrumentation, direct comparisons with other researchers are
difficult. Hager and coworkers reported a detection limit of 0.006 ppm for Sr in a glass
matrix, but their work was performed with an IR laser with much higher energy

Ni Intensity (counts/s)
5x10
4x10
3x10
2x10
1x10
Concentration (ppm)
Figure 8-4. Calibration plot for Ni in copper samples.

Normalized Sr Intensity
2.5
Concentration (ppm)
Figure 8-5. Calibration plot for Sr in glass samples.
4^

146
(hundreds of millijoules) and a higher repetition rate (10 Hz) (12). This would result in
much larger quantities being introduced to the ICP-MS and therefore, at least partially,
explains the much lower detection limit. In addition, it is not clear whether single ion
monitoring was used (as opposed to multielement scanning) in determining the detection
limit, as this can result in significant improvements in detection limits.
Conclusions
The results obtained with the compact laser system indicate that it can be
successfully coupled with an ICP-MS to achieve sub-ppm detection limits for analytes in
both conducting and nonconducting matrices. Detection limits for Sr in the NIST glasses
are a factor of two higher than those obtained with the commercially available UV laser
system. In addition, spatial resolution is not as good for IR lasers compared to UV lasers.
This is due to diffraction limitations, as well as differences in laser sampling mechanisms
for IR and UV lasers (laser-material, laser-plasma, and plasma-material interactions). For
bulk sampling of solids, however, the compact Nd:YAG provides a potentially useful,
simple, and relatively inexpensive option for ICP-MS users.

CHAPTER 9
CONCLUSIONS AND FUTURE WORK
Conclusions
Laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) has
been used to analyze a variety of materials. The technique has a number of advantages,
most notable is the ability to directly analyze essentially any solid material with little or
no sample preparation. It is a sensitive (sub-ppm detection limits) multielement
technique that can be used for both bulk and low-resolution (pm scale) spatial analysis.
As with most other solids techniques, the major limitation to the widespread applicability
of LA-ICP-MS is the existence of matrix effects. The matrix effects are relatively minor,
however, and semiquantitative work is easily performed. For quantitative analysis,
matrix-matched standards are usually required. Unfortunately for some materials,
standards do not exist and alternative calibration approaches must be sought.
In this work, the use of solutions for calibration of solid materials was studied. It
was demonstrated that reasonably accurate results (± 20 % typically) could be obtained
from a range of samples (glass, ceramic, soil, and coal fly ash), provided that proper
ablation and ICP-MS conditions were chosen. An additional benefit of using the “wet
plasma” configuration associated with dual sample introduction was that it provided
higher sensitivity and better precision. The importance of laser irradiance and sampling
strategy on representative sampling was demonstrated for glasses. The laser irradiance
147

148
o /y
should exceed 2x10 W/cm to avoid fractionation of elements during the ablation
process. Sampling strategy (stationary of translated sample) was an important
consideration for all analyses. Translation of the sample resulted in larger ablation rates;
however, the sampling depth was considerably smaller than depths produced without
translation. The sample heterogeneity dictated which sampling strategy was most
suitable for a particular material.
Solution calibration was extended to the analysis of ceramics. Silicon nitride
ceramic bearings were analyzed and the results for minor elements obtained with the
solution calibration approach were confirmed by X-ray microanalysis. In addition, glass
standard reference materials were demonstrated to be useful for the analysis of silicon
nitride ceramics, provided the difference in ablated mass between the samples was
accounted for. This led to an investigation of light scattering for mass normalization.
Mass normalization based on light scattering was useful. It effectively corrected
for differences in ablated mass between a glass and a ceramic material, as a consequence
of similar-sized ablated particulates from these materials. This meant that absolute
sensitivity values determined from a standard glass material, rather than relative
sensitivity factors, could be used for analyzing ceramics. In other words, the
concentration of an element in the sample of interest (internal standard) was not required.
Use of an internal standard of known concentration was required with solution
calibration. An internal standard was also used when calibrating with solids, since the
ablated mass typically varied from sample to sample. Using light scattering eliminated
restrictions associated with internal standardization, such as the requirement that the

149
analyte and internal standard be distributed similarly in the solid. Light scattering might
be most useful for analyzing inhomogeneous or multilayered samples.
The analysis of soils presented several challenges that were unique to this matrix.
The importance of selecting an appropriate internal standard was demonstrated for the
analysis of soils using solution calibration. Interferences on Ca isotopes prevented their
use as internal standards for quantitative work; therefore, a trace element (Ni) was used to
demonstrate the utility of solution calibration. An internal standard that behaved similar
to the analyte with regards to ICP-MS matrix effects was required. This was important
for the analysis of Sr, Y, Rb, and Ba in soil, since the relative signals for these analytes
were suppressed when compared to aqueous solutions. The effect of element speciation
on LA-ICP-MS signals was studied, and it was demonstrated that signals were largely
unaffected by the anionic constituent of the analyte. In addition, the level of organic
material in the sample had no appreciable affect on LA-ICP-MS results. For pressed
particulate samples, particle size of the matrix constituents significantly influenced the
precision of measurements. Precision improved as the particle size was reduced. Particle
size was demonstrated to be an important consideration with standard additions
measurements as well, since relative analyte signals for spiked sand samples increased
when the particle size exceeded 35 pm.
Finally, a compact Nd:YAG laser was evaluated for the analysis of glass and
metal samples. It was demonstrated that the laser might be useful for bulk analysis of
solids, since it was inexpensive, easily set-up, and provided sub ppm detection limits for
analytes in a variety of matrices.

150
Future Work
This research has opened up a range of future investigations. Light scatter
normalization should be refined and extended to determine the full extent of its
capabilities. The influence of scatter collection angle on the effectiveness of
normalization within and between matrices should be investigated. In addition, the work
should be extended to applications where normal internal standardization cannot be used,
such as with multilayered ceramics or inhomogeneous geological samples. In these
cases, analyte signals could be measured with the ICP-MS and the light scatter signal
could be used for mass correction. Therefore, the concentration of the analyte throughout
the sample could be reliably determined. This would not be possible with internal
standardization, since no element would be present throughout the sample at identical
levels. Light scattering, however, does depend on the physical properties (refractive
index, particle size, etc.) of the ablated material, and the scatter signal-mass signal
relationship will have to be explored more thoroughly.
The spatial and depth profiling capabilities of the LA-ICP-MS system need to be
examined. Single shot ablation removes sub-pm layers from the sample surface,
therefore, the analysis of pm layers of coatings is possible. Additionally, LA can be used
to analyze relatively thick layers (hundreds of microns), which cannot be done with other
surface analysis techniques. Application of LA-ICP-AES to the analysis of thick layers
of coating has recently been reported (16); however, the use of LA-ICP-MS could be
advantageous if high sensitivity is required. The in-depth resolution of laser sampling is
determined by the rate at which material is removed from the surface; therefore, low
repetition rates and/or translation of the sample should be employed. Low repetition rates

151
can be problematic, however, because transient signals are generated. With a scanning
instrument, this makes mass normalization difficult, since the signal is changing rapidly
in comparison to the time spent at each mass location. If light scatter normalization is
used, the scattering signal that is produced from a single pulse can be used for mass
correction. Light scatter normalization based on single shot measurements has not been
examined in this work, however, scattering signals have been obtained for single shots on
glass samples.
Additional work could be performed with soil and particulate samples. The lower
relative sensitivities for Sr, Rb, Y, and Ba were explained on the basis of high levels of
easily ionizable elements (EIE). The EIE effect could be studied by spiking these
analytes, as well as several others analytes for which no suppression was observed, into
several different matrices, such as graphite and sand. To these matrices, increasing levels
of an EIE (e. g. Na) could be added to determine if the EIE affected relative analyte
signals. Also, more extensive studies on ICP-MS operating conditions would be useful to
determine if suitable conditions could be obtained for the analysis of all analytes in soil.
In a study involving solution nebulization ICP-MS of dissolved steel samples, it was
demonstrated that nebulizer flow rate significantly influenced the extent of matrix effects
(102). By decreasing the nebulizer flow rate, matrix effects were reduced, although with
a loss in sensitivity. Similarly, the influence of flow conditions on matrix effects in LA-
ICP-MS analyses could be studied.
Additionally, the combined use of standard additions and solution calibration
could be explored. As mentioned, the major limitation of solution calibration was that
the concentration of the internal standard was required. For the analysis of glass and

152
ceramics, this limitation could be overcome by measuring light scattered from the
ablation product. For soil and particulate samples, an internal standard of known
concentration could be spiked directly into the sample. A brief survey scan of the sample
of interest could be used to identify elements that did not exist at appreciable levels in the
sample. This element could then be spiked into the sample, to serve as the internal
standard. As demonstrated in Chapter 7, the applicability of spiked samples for
calibration would depend on the type and size of particles present in the sample. The use
of internal standards could be studied for a wide range of materials (soils, ceramics
powders, etc.) and analytes to determine the limitations of this approach.

REFERENCES
1. M. Franek, V. Krivan, B. Gercken, and J. Pavel, Mikrochim. Acta 113, 251
(1994).
2. B. Gercken, J. Pavel, and O. Suter, in Applications of Plasma Source Mass
Spectrometry II, G. Holland and A. N. Eaton, Eds. (The Royal Society of
Chemistry, Thomas Graham House, Cambridge, 1993).
3. V. Balaram, Current Sci. 69, 640 (1995).
4. K. H. Wedepohl, Ed. Handbook of Geochemistry, Vol. I-II (Springer-Verlag,
New York, 1970).
5. G. Zaray and T. Kantor, Spectrochim. Acta 50B, 489 (1995).
6. A. S. Eppler, D. A. Cremers, D. D. Hickmott, M. J. Ferris, and A. C. Koskelo,
Appl. Spectrosc. 50, 1175 (1996).
7. J. C. Van Loon, Selected Methods of Trace Metal Analysis: Biological and
Environmental Samples (John Wiley and Sons, New York, 1985).
8. R. Jenkins, X-Ray Fluorescence Spectrometry (John Wiley and Sons, New York,
1988).
9. J. I. Goldstein, D. E. Newbury, P. Echlin, D. C. Joy, A. D. Romig, Jr., C. E.
Lyman, C. Fiori, and E. Lifshin, Scanning Electron Microscopy and X-Ray
Microanalysis (Plenum Press, New York, 1994).
10. A. Lodding, in Inorganic Mass Spectrometry, F. Adams, R. Gijbels, and R. Van
Grieken, Eds. (John Wiley and Sons, New York, 1988).
11. W. W. Harrison, in Inorganic Mass Spectrometry, F. Adams, R. Gijbels, and R.
Van Grieken, Eds. (John Wiley and Sons, New York, 1988).
12. E. R. Denoyer, K. J. Fredeen, and J. W. Hager, Anal. Chem. 63, 445A (1991).
13. D. M. Hercules and S. H. Hercules, J. Chem Ed. 61, 592 (1984).
14. D. Milton and J. C. Hutton, Spectrochim. Acta 48B, 39 (1993).
153

154
15. R. K. Marcus, J. Anal. At. Spectrom. 11, 821 (1996).
16. V. Kanicky, I. Novotny, J. Musil, and J. -M. Mermet, Appl. Spectrosc. 51, 1042
(1997).
17. S. A. Darke and J. F. Tyson, J. Anal. At. Spectrom. 8, 145 (1993).
18. R. E. Russo, Appl. Spectrosc. 49, 14A (1995).
19. A. A. van Heuzen, Spectrochim. Acta 46B, 1803 (1991).
20. W. T. Perkins, R. Fuge, and N. J. G. Pearce, J. Anal. At. Spectrom. 6, 445 (1991).
21. A. A. van Heuzen and J. B. W. Morsink, Spectrochim. Acta 46B, 1819 (1991).
22. Inductively Coupled Plasmas in Analytical Atomic Spectrometry, Second Edition,
A. Montaser and D. W. Golightly, Eds. (VCH Publishers, New York, 1992).
23. Handbook of Inductively Coupled Plasma Mass Spectrometry, K. E. Jarvis, A. L.
Gray, and R. S. Houk, Eds. (Chapman and Hall, New York, 1992).
24. D. W. Douglas, Prog. Analyt. Atom. Spectrosc. 8, 1 (1995).
25. R. S. Houk, Anal. Chem. 58, 97A (1986).
26. J. W. Olesik, Anal. Chem. 63, 12A (1991).
27. M. W. Blades and D. G. Weir, Spectroscopy, 9, 14 (1994).
28. D. J. Douglas and J. B. French, J. Anal. At. Spectrom. 3, 743 (1988).
29. SOLA ICP/GD Mass Spectrometer Operators Manual, Finnigan MAT Ltd. San
Jose, 1993.
30. F. Vanhaecke, L. Moens, R. Dams, I. Papadakis, and P. Taylor, Anal. Chem. 69,
268 (1997).
31. P. P. Mahoney, S. J. Ray, and G. M. Hieftje, Appl. Spectrosc. 8, 16A, (1997).
32. D. W. Koppenaal, C. J. Barinaga, and M. R. Smith, J. Anal. At. Spectrom. 9, 1053
(1994).
33. R. C. Hutton and A. N. Eaton, J. Anal. At. Spectrom. 2, 595 (1987).
34. D. G. Weir and M. W. Blades, Spectrochim. Acta 45B, 615 (1990).

155
35. J. W. McLaren, J. W. Lam, and A. Gustavsson, Spectrochim. Acta 45B, 1091
(1990).
36. R. Tsukahara and M. Kubota, Spectrochim. Acta 45B, 581 (1990).
37. A. L. Gray, Analyst 110, 551 (1985).
38. M. Thompson, S. Chenery, and L. Brett, J. Anal. Atom. Spectrom. 4, 11 (1989).
39. V. V. Kogan, M. W. Hinds, and G. I. Ramendik, Spectrochim. Acta 49B, 333
(1994).
40. J. L. Imbert and P. Telouk, Mikrochim. Acta 110, 151 (1990).
41. J. S. Becker, U. Breuer, J. Westheide, A. I. Saprykin, H. Holzbrecher, H. Nickel,
and H. -J. Dietze, Fres. J. Anal. Chem. 355, 626 (1996).
42. I. D. Abell, D. Gregson, and S. Shuttleworth, in The Physics and Chemistry of
Carbides, Nitrides, and Borides, R. Freer, Ed. (Kluwer Academic Publishers,
Dordrecht, The Netherlands, 1990).
43. F. E. Lichte, Anal. Chem. 67, 2479 (1995).
44. A. M. Ghazi, T. E. McCandless, D. A. Vanko, and J. Ruiz, J. Anal. At. Spectrom.
11, 667(1996).
45. P. M. Outridge and R. D. Evans, J. Anal. At. Spectrom. 10, 595 (1995).
46. J. Marshall, F. Franks, I. Abell, and C. Tye, J. Anal. At. Spectrom. 6, 145 (1991).
47. N. Bloembergen, in Laser Solid Interactions and Laser Processing, S. D. Ferris,
H. J. Leamy, and J. M. Poate, Eds. (American Institute of Physics, New York,
1979).
48. J. F. Ready, Effect of High-Powered Laser Radiation (Academic Press, New
York, 1971).
49. R. Wisbrun, I. Schechter, R. Niessner, H. Schroder, and K. L. Kompa, Anal.
Chem. 66, 2964 (1994).
50. F. Hillenkamp, M. Karas, R. Beavis, and B. Chait, Anal. Chem. 63, 1193A (1991).
51. W. Sdorra, A. Quentmeier, and K. Niemax, Mikrochim. Acta II, 185 (1989).

156
52. T. Kantor, L. Bezur, E. Pungor, and P. Fodor, Spectrochim. Acta 34B, 341
(1979).
53. R. Wennrich and K. Dittrich, Spectrochim. Acta 42B, 995 (1987).
54. Y. Iida, Spectrochim. Acta 45B, 427 (1990).
55. A. Ciocan, L. Hiddemann, J. Uebbing, and K. Niemax, J. Anal. At. Spectrom. 8,
273 (1993).
56. Y. Q. Tang and C. Trassy, Spectrochim. Acta 41B, 143 (1986).
57. S. E. Long and R. F. Browner, Spectrochim. Acta 43B, 1461 (1988).
58. C. N. Davies, Aerosol Science (Academic Press, New York, 1966).
59. P. Arrowsmith and S. K. Hughes, Appl. Spectrosc. 42, 1231 (1988).
60. L. Ebdon, M. E. Foulkes, and S. Hill, J. Anal. At. Spectrom. 5, 67 (1990).
61. K. E. Jarvis and J. G. Williams, Chem. Geol. 77, 53 (1989).
62. R. J. Watling, B. F. Lynch, and D. Herring, J. Anal. At. Spectrom. 12, 195 (1997).
63. A. Moissette, T. J. Shepherd, and S. R. Chenery, J. Anal. At. Spectrom. 11, 177
(1996).
64. J. Stix, G. Gauthier, and J. N. Ludden, Can. Mineral. 33, 435 (1995).
65. Z. Chen, W. Doherty, and D. C. Gregoire, J. Anal. At. Spectrom. 12, 653 (1997).
66. B. J. Fryer, S. E. Jackson, and H. P. Longerich, Can. Mineral. 33, 303 (1995).
67. L. Moenke-Blankenburg, T. Schumann, D. Günther, H. -M. Kuss, and M. Paul, J.
Anal. At. Spectrom. 7, 251 (1992).
68. M. Ducreux-Zappa and J. -M. Mermet, Spectrochim. Acta 51B, 321 (1996).
69. E. F. Cromwell and P. Arrowsmith, Anal. Chem. 67, 131 (1995).
70. S. Chenery and J. M. Cook, J. Anal. At. Spectrom. 8, 299 (1993).
71. J. W. Hager, Anal. Chem. 61, 1243 (1989).
72. E. F. Cromwell and P. Arrowsmith, Appl. Spectrosc. 49, 1652 (1995).

157
73. H. P. Longerich, D. Günther, and S. E. Jackson, Fres. J. Anal. Chem. 355, 538
(1996).
74. X. Mao, W. -T. Chan, M. Caetano, M. A. Shannon, and R. E. Russo, Appl.
Surface Sci. 96-98, 126 (1996).
75. P. M. Outridge, W. Doherty, and D. C. Gregoire, Spectrochim. Acta 51B, 1451
(1996).
76. T. E. Jeffries, N. J. G. Pearce, W. T. Perkins, and A. Raith, Anal. Comm. 33, 35
(1996).
77. C. Geertsen, A. Briand, F. Chartier, J. -L. Lacour, P. Mauchien, S. Sjóstróm, and
J. -M. Mermet, J. Anal. At. Spectrom. 9, 17 (1994).
78. D. Figg and M. S. Kahr, Appl. Spectrosc. 51, 1185 (1997).
79. CRC Handbook of Chemistry and Physics, D. R. Lide, Ed. (CRC Press, Boca
Raton, 1994).
80. J. F. Chudecki, Ceramic Bulletin 69, 1113 (1990).
81. J. A. C. Broekaert and G. Tólg, Mikrochim. Acta II, 173 (1990).
82. J. A. C. Broekaert, T. Graule, H. Jenett, G. Tólg, and P. Tschópel, Fres. J. Anal.
Chem. 332, 825 (1989).
83. L. K. L. Falk and E. U. Engstróm, J. Am. Ceram. Soc. 74, 2286 (1991).
84. C. Ziegler, J. Heinrich, and G. Wotting, J. Mat. Sci. 22, 3041 (1987).
85. E. H. Homeier, S. A. Bradley, and K. R. Karasek, J. Mat. Sci. 27, 1231 (1992).
86. J. S. Crighton, J. Carroll, B. Fairman, J. Haines, and M. Hinds, J. Anal. At.
Spectrom. 11, 461R (1996).
87. J. T. Westheide, J. S. Becker, R. Jáger, H. -J. Dietze, and J. A. C. Broekaert, J.
Anal. At. Spectrom. 11, 661 (1996).
88. P. Arrowsmith, in Ceramic Transactions Vol. 5, W. S. Young, G. L. McVay, and
G. Pike, Eds. (American Ceramic Society, Westerville, OH, 1988).
89. H. Pang, D. R. Wiederin, R. S. Houk, and E. S. Yeung, Anal. Chem. 63, 390
(1991).

158
90. T. Tanaka, K. Yamamoto, T. Nomizu, and H. Kawaguchi, Anal. Sci. 11, 967
(1995).
91. P. Richner, M. W. Borer, K. R. Brushwyler, and G. M. Hieftje, Appl Spectrosc.
44, 1290 (1990).
92. M. Ducreux-Zappa and J. -M. Mermet, Spectrochim. Acta 51B, 333 (1996).
93. M. Thompson, S. Chenery, and L. Brett, J. Anal. At. Spectrom. 5, 49 (1990).
94. Y. Huang, Y. Shibata, and M. Morita, Anal. Chem. 65, 2999 (1993).
95. K. Y. Yamamoto, D. A. Cremers, M. J. Ferris, and L. E. Foster, Appl. Spectrosc.
50, 222 (1996).
96. R. Barbini, F. Colao, R. Fantoni, A. Palucci, H. J. L. van der Steen, and M.
Angelone, Appl. Phys. B 65, 101 (1997).
97. S. F. Durrant and N. I. Ward, Fres. J. Anal. Chem. 345, 512 (1993).
98. L. Moenke-Blankenburg, T. Schumann, and J. Nólte, J. Anal. At. Spectrom. 9,
1059(1994).
99. X. Guo and F. E. Lichte, Analyst 120, 2707 (1995).
100. E. Hoffmann, C. Liidke, and H. Stephanowitz, Fres. J. Anal. Chem. 355, 900
(1996).
101. J. W. Olesik, Anal. Chem. 68, 469A (1996).
102. M. -A. Vaughan and G. Horlick, J. Anal. At. Spectrom. 4, 45 (1989).
103. P. Allain, L. Jaunault, Y. Mauras, J. -M. Mermet, and T. Delaporte, Anal. Chem.
63, 1497(1991).
104. H. P. Longerich, J. Anal. At. Spectrom. 4, 665 (1989).
105. Kigre, Inc. 100 Marshland Road, Hilton Head Island, SC 29926.

BIOGRAPHICAL SKETCH
Scott A. Baker was born in Pensacola, Florida, on June 10, 1971. He grew up in
Charleston, South Carolina, and graduated from James Island High School in 1989. He
received his Bachelor of Science degree in chemistry in May 1993 from the College of
Charleston. He started graduate school at the University of Florida in August 1993,
where he joined Dr. James D. Winefordner’s research group. He received his Doctor of
Philosophy in analytical chemistry in May 1998.
159

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
James D. Winefordner, Chairman
Graduate Research Professor of Chemistry
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Willard W. Harrison
Professor of Chemistry
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Robert T. Kennedy
Associate Professor of <
emistry
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
kl k Ljh- â– .
William Weltner
Professor of Chemistry

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Ellsworth D. Whitney
Professor of Materials Science
ngineering
This dissertation was submitted to the Graduate Faculty of the Department of
Chemistry in the College of Liberal Arts and Sciences and to the Graduate School and
was accepted as partial fulfillment of the requirements for the degree of Doctor of
Philosophy.
May 1998
Dean, Graduate School

UNIVERSITY OF FLORIDA
3 1262 08555 1066



xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EAUTJ7OWS_K6UCIB INGEST_TIME 2014-07-22T21:16:01Z PACKAGE AA00018886_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES