<%BANNER%>

Laser-Induced Plasmas as an Analytical Source for Quantitative Analysis of Gaseous and Aerosol Systems

Permanent Link: http://ufdc.ufl.edu/UFE0024940/00001

Material Information

Title: Laser-Induced Plasmas as an Analytical Source for Quantitative Analysis of Gaseous and Aerosol Systems Fundamentals of Plasma-Particle Interactions
Physical Description: 1 online resource (161 p.)
Language: english
Creator: Diwakar, Prasoon
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: analyte, diffusion, laser, libs, matrix, plasma, scattering, spectroscopy, thomson
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Laser-induced Breakdown Spectroscopy (LIBS) is a relatively new analytical diagnostic technique which has gained serious attention in recent past due to its simplicity, robustness, and portability and multi-element analysis capabilities. LIBS has been used successfully for analysis of elements in different media including solids, liquids and gases. Since 1963, when the first breakdown study was reported, to 1983, when the first LIBS experiments were reported, the technique has come a long way, but the majority of fundamental understanding of the processes that occur has taken place in last few years, which has propelled LIBS in the direction of being a well established analytical technique. This study, which mostly focuses on LIBS involving aerosols, has been able to unravel some of the mysteries and provide knowledge that will be valuable to LIBS community as a whole. LIBS processes can be broken down to three basic steps, namely, plasma formation, analyte introduction, and plasma-analyte interactions. In this study, these three steps have been investigated in laser-induced plasma, focusing mainly on the plasma-particle interactions. Understanding plasma-particle interactions and the fundamental processes involved is important in advancing laser-induced breakdown spectroscopy as a reliable and accurate analytical technique. Critical understanding of plasma-particle interactions includes study of the plasma evolution, analyte atomization, and the particle dissociation and diffusion. In this dissertation, temporal and spatial studies have been done to understand the fundamentals of the LIBS processes including the breakdown of gases by the laser pulse, plasma inception mechanisms, plasma evolution, analyte introduction and plasma-particle interactions and their influence on LIBS signal. Spectral measurements were performed in a laser-induced plasma and the results reveal localized perturbations in the plasma properties in the vicinity of the analyte species, for first 60 ?s. The measurements provide direct evidence of matrix effects in the LIBS plasma at early times. Electron density measurements at very early times in the plasma reveal deviations from the Local Thermodynamic Equilibrium conditions (LTE). The data from various experiments suggests a complex interaction between the plasma and the aerosol particle, during which the finite time-scales of particle dissociation, and heat and mass transfer are fundamental processes.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Prasoon Diwakar.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Hahn, David W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024940:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024940/00001

Material Information

Title: Laser-Induced Plasmas as an Analytical Source for Quantitative Analysis of Gaseous and Aerosol Systems Fundamentals of Plasma-Particle Interactions
Physical Description: 1 online resource (161 p.)
Language: english
Creator: Diwakar, Prasoon
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: analyte, diffusion, laser, libs, matrix, plasma, scattering, spectroscopy, thomson
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Laser-induced Breakdown Spectroscopy (LIBS) is a relatively new analytical diagnostic technique which has gained serious attention in recent past due to its simplicity, robustness, and portability and multi-element analysis capabilities. LIBS has been used successfully for analysis of elements in different media including solids, liquids and gases. Since 1963, when the first breakdown study was reported, to 1983, when the first LIBS experiments were reported, the technique has come a long way, but the majority of fundamental understanding of the processes that occur has taken place in last few years, which has propelled LIBS in the direction of being a well established analytical technique. This study, which mostly focuses on LIBS involving aerosols, has been able to unravel some of the mysteries and provide knowledge that will be valuable to LIBS community as a whole. LIBS processes can be broken down to three basic steps, namely, plasma formation, analyte introduction, and plasma-analyte interactions. In this study, these three steps have been investigated in laser-induced plasma, focusing mainly on the plasma-particle interactions. Understanding plasma-particle interactions and the fundamental processes involved is important in advancing laser-induced breakdown spectroscopy as a reliable and accurate analytical technique. Critical understanding of plasma-particle interactions includes study of the plasma evolution, analyte atomization, and the particle dissociation and diffusion. In this dissertation, temporal and spatial studies have been done to understand the fundamentals of the LIBS processes including the breakdown of gases by the laser pulse, plasma inception mechanisms, plasma evolution, analyte introduction and plasma-particle interactions and their influence on LIBS signal. Spectral measurements were performed in a laser-induced plasma and the results reveal localized perturbations in the plasma properties in the vicinity of the analyte species, for first 60 ?s. The measurements provide direct evidence of matrix effects in the LIBS plasma at early times. Electron density measurements at very early times in the plasma reveal deviations from the Local Thermodynamic Equilibrium conditions (LTE). The data from various experiments suggests a complex interaction between the plasma and the aerosol particle, during which the finite time-scales of particle dissociation, and heat and mass transfer are fundamental processes.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Prasoon Diwakar.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Hahn, David W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024940:00001


This item has the following downloads:


Full Text

PAGE 1

1 LASER INDUCED PLASMAS AS AN ANALYTICAL SOURCE FOR QUANTITATIVE ANALYSIS OF GASEOUS AND AEROSOL SYSTEMS: FUNDAMENTALS OF PLASMAPARTICLE INTERACTIONS By PRASOON K DIWAKAR A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF T HE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Prasoon K Diwakar

PAGE 3

3 To my parents

PAGE 4

4 ACKNOWLEDGMENTS I would lik e to acknowledge my sincere thanks to Dr. David Hahn for his guidance and support throug hout my PhD program His enthusiasm, immense knowledge of the subject and constant encouragement provided me motivation and th e opportunity to learn a lot during my tim e at the University of Florida. I would also like to thank him for his patience and support during troubled times in my personal life. As a mentor, ad visor, and teacher he has provided me confidence and stimulated my interest to pursue my career in the field of academic research. I would also like to thank my lab mates (Allen, Bret, Benoit, Brian, Cary, Kibum, Katie, Leia, Michael, Phil, Soupy and Vince) for their help and assistantship in my research work as well as for all th e insightful discussion sessio ns which made the lab a fun place to work Last but not the least, I would like to thank my family, my girlfriend, Mitheila, for all the love, support and constant encouragement to do my best in career and life.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................................................................................................... 4 LIST OF TABLES ............................................................................................................................... 7 LIST OF FIGURES .............................................................................................................................. 8 ABSTRACT ........................................................................................................................................ 12 CHAPTER 1 LIBS ............................................................................................................................................. 14 Introduction ................................................................................................................................. 14 LIBS in Differ ent Media: Solids, Liquids, Gases ..................................................................... 15 Why LIBS? .................................................................................................................................. 16 Basics of LIBS ............................................................................................................................ 18 Temporal Nature of LIBS Signal ............................................................................................... 19 Laser Induced Plasma ................................................................................................................. 20 Transient Plasma Dynamics ....................................................................................................... 21 Fractionation and Matrix Effect ................................................................................................. 23 Transient Electron Density ......................................................................................................... 26 Stark Broadening Method ................................................................................................... 26 Saha Boltzmann Method ..................................................................................................... 28 Langmuir Probe ................................................................................................................... 30 Single Droplet Study ................................................................................................................... 30 Motivation and Objectives ......................................................................................................... 32 2 AEROSOL SAMPLING POISSON DISTRIBUTION ....................................................... 34 Introduction ................................................................................................................................. 34 Poisson Distribution .................................................................................................................... 37 Sahara Dust Storm Experiment .................................................................................................. 38 Experimental Setup ..................................................................................................................... 39 Results and Discussions .............................................................................................................. 42 3 FRACTIONATION AND MATRIX EFFECT ........................................................................ 49 Introduction ................................................................................................................................. 49 Experimental Methods ................................................................................................................ 50 Aerosol Generation ..................................................................................................................... 51 Results and Discussion ............................................................................................................... 52 Analyte Emission Enhancement with Added Aerosol Mass ............................................ 52 Localized Plasm a Measurements for Mg/Cd Aerosol Particles ....................................... 61

PAGE 6

6 4 ELECTRON DENSITY MEASUREMENTS .......................................................................... 73 Introduction ............................................................................................................................... 73 Thomson Scattering .................................................................................................................... 73 Thomson Scattering Theory ....................................................................................................... 75 Experimental Setup ..................................................................................................................... 84 Thomson Scattering Imaging ..................................................................................................... 88 N Line Stark Broadening ............................................................................................................ 95 Interpretation of Thomson Res ults and Electron Density Measurements ............................. 100 Transmission Experiments ....................................................................................................... 102 Investigation of Laser Breakdown in Gases ............................................................................ 104 Conclusion ................................................................................................................................. 111 5 SINGLE DROPLET INVESTIGATION IN LIBS ................................................................. 113 Introduction ............................................................................................................................... 113 Experimental Setup ................................................................................................................... 115 Si ngle Droplet Sample Introduction S ystem ................................................................... 115 Optical and Data Collection S etup ................................................................................... 118 Calibration Curves and Limits of Detection for Ca and Au ................................................... 120 Local Temperature Measu rements ........................................................................................... 125 Diffusion Coefficient Measurements ....................................................................................... 132 Conclusions ............................................................................................................................... 138 6 CONCLUSION AND FUTURE WORK ................................................................................ 140 Plasma Inception Study ............................................................................................................ 141 Electron Density Measurements .............................................................................................. 142 Fractionation and Matrix Effects Study................................................................................... 144 Investigation of the Single Droplet in LIBS Plasma .............................................................. 147 Aerosol Sampling Study ........................................................................................................... 149 Future Work .............................................................................................................................. 151 REFERENCES ................................................................................................................................. 1 53 BIO GRAPHICAL SKETCH ........................................................................................................... 161

PAGE 7

7 LIST OF TABLES Table page 2 1 Experimental and Poisson sampling probabilities ............................................................... 45 3 1 Turbidity values for Cd transitions for different delays ...................................................... 66 3 2 Turbidity values for Mg transitions for different delays .................................................... 66 3 3 Properties of Mg and Cd used in calculation of ion to neutral ratio ................................... 67 4 1 Spectral properties of N(II) lines .......................................................................................... 96 4 2 Stark broadening parameter for Nitrogen ............................................................................. 97 4 3 Griems parameter calculated for different temperatures by curve fitting ......................... 98 4 4 Electron density as a function of delay time ........................................................................ 99 5 1 Droplet generator parameters .............................................................................................. 117 5 2 Calcium detection limits in various LIBS studies for isolated drop lets and aqueous samples .................................................................................................................. 124 5 3 Lu spectroscopic data .......................................................................................................... 129 5 4 Calcium, Hydrogen and Nitr ogen properties for calculation o f diffusion coefficients ........................................................................................................................... 133 5 5 Lennard Jones parameters for Ca ion and H atom attached to water ............................... 134 5 6 Hydrogen Diffusion coefficient calculations ..................................................................... 136

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Typical LIBS experim ental setup ........................................................................................ 18 1 2 Temporal nature of the LIBS signal ..................................................................................... 19 1 3 ne and fitted Voigt profile ............................................ 27 1 4 Typic al Optical trapping set up. ........................................................................................... 31 2 1 Effect of Dust con centration on coral ecolo gy in Florida and Caribbean ......................... 39 2 2 Experiment setup for ambient air study ............................................................................... 40 2 3 LIBs spectra recorde d in ambient air .................................................................................... 43 2 4 Dust concentration during 25th Jul 29th Jul 2005 ................................................................. 44 2 5 Sampling probability of aerosol in LIBS .............................................................................. 46 2 6 Comparison of single shot spectra to ensemble average of 20,000 shots .......................... 47 3 1 Top view of experimental apparatus for t he LIBS aero sol experiments [ ......................... 50 3 2 A erosol generation system ................................................................................................... 51 3 3 TEM image showing Cd: Zn particles generated by nebulizing and drying 1:4 mass ratio solution. .......................................................................................................... 53 3 4 Sodium emission spectra recorded for pure sodium -based aerosols and for sodium -copper containing aerosols at a 1:9 mass ratio. ...................................................... 55 3 5 Enhancement of the sodium doublet emission intensity as a function of delay time with respect to plasma initiation for sodium -copper, sodium -zinc, and sodium -tungsten containing aerosols at a 1:9 mass ratio. ................................................... 56 3 6 Sodium atomic emission signal in the presence of added copper normalized to sodium -only emission. ....................................................................................................... 57 3 7 Enhancement of the magne sium neutral emission intensity (285.2 nm) as a function of delay time with respect to plasma initiation for magnesium -zinc, and magnesium tungsten containing aerosols at a 1:19 mass ratio. ................................... 58 3 8 Magnesium emission spectra recorded for magnesium -cadmium aerosol particles at a 1:17 mass ratio as a function of delay time.. .................................................. 62

PAGE 9

9 3 9 Cadmium emission spectra recorded for magnesium -cadmi um aerosol particles at a fixed delay of 25 ............................. 63 3 10 Cadmium ion-to -neutral (226.5 to 228.8 nm) intensity ratios as a function of delay time with respect to plasma initiation for magnesium -cadmium mass ratios of 1:5, 1:11, and 1:17. .................................................................................................. 64 3 11 Ratio of the magnesium ion-to -neutral (280.27 and 285.21 nm) intensity ratio to the cadmium ion to neutral (226.5to 228.8 nm) intensity ratio as a f unction of plasma temperature based on the theoretical tr eatment of Equation 3.1 .......................................................................................................................... 67 3 12 Calculated plasma temperatures based on the magnesium to -cadmium intensity ratios as a function of del ay time with respect to plasma initiation based on the theoretical treatment of Equation 3.11 .......................................................... 68 4 1 Thomson scattering geometry .............................................................................................. 75 4 2 Re gimes in Thomson scattering ........................................................................................... 77 4 3 ............................................................ 77 4 4 Coherent and In coherent Thomson scattering ..................................................................... 79 4 5 Top view schematic of the experimental setup for plasma scattering imaging and transmission experiments. .............................................................................................. 85 4 6 Line width of 532 nm line filter ............................................................................................ 86 4 7 Waveforms of the LIBS laser pulse and the probe laser for a peak to peak temporal delay of 50 ns. The 20 -ns intensifier gate width is synchronized to the probe laser, as shown. ...................................................................................................... 87 4 8 Ty pical Thomson scattering imaging. Grid size is also shown here which yields resolution of camera to be ~7.8m /pixel. .................................................................. 89 4 9 Raw images of the plasma only (above) and of the plasma with scatter ing probe (below) at different delay times with respect to plasma creating laser. ................... 89 4 10 Processed plasma scattering images, in which the background plasma images have been subtracted from the plasma plus probe laser images. ............................ 90 4 11 Temporal evolution of Thomson scattering as a function of delay time. ........................... 91 4 12 Spectral scattering profile of the probe laser for the high energy LIBS laser recorded at a delay of 25 ns ................................................................................................... 92 4 13 Close up of the Thomson scattering spectrum presented in Figure 4 12 ........................... 94

PAGE 10

10 4 14 Further zoomed Thomson scattering spectrum .................................................................... 94 4 15 N(II) line broadening and fitted Lorentzian profile ............................................................. 95 4 16 Lorentzian fitted profile of three N lines .............................................................................. 96 4 17 Instrument profile correction using measured line width of Hg lamp ................................ 97 4 18 Griems parameter plotted against Temperature. Third degree polynomial is fitted to obtain values of C at intermediate temperatures. ................................................... 98 4 19 Transient elec tron density at 25,000 K calculated via Stark broadening ......................... 100 4 20 Thomson scattering parameter for different electron densities ......................................... 101 4 21 Measured transmission of the probe laser beam through the plasma as a function of delay time for high energy LIBS laser. ........................................................... 104 4 22 Energy deposition in laser induced breakdown by laser s having peak energies 1 and 2.5 times the breakdown threshold energy .............................................. 106 4 23 Temporal laser pulse profile used in current study ............................................................ 108 4 24 Breakdown images for Nitrogen, Argon, Helium at delay times of 0, 1, 10 and 20 ns after the plasma inception. Laser direction is from left to right. ...................... 109 4 25 Lines of constant int ensity, isophytes plotted for optical systems with (a) no aberration and (b) high degree of aberration ..................................................................... 109 5 1 50 m single droplet of analyte generated using a scientific grade droplet gener ator ............................................................................................................................... 117 5 2 Schematic of experimental setup for single droplet experiments ..................................... 119 5 3 Front view of the schematic of experimental setup for single droplet experiments .......................................................................................................................... 119 5 4 Representative Au spectra showing range of LIBS response for concentration of 1000 g/ml ............................................................................................... 121 5 5 Histogram of P/B for Au at 267.59 nm for concentrations of 100, 500 and 1000 g/ml ........................................................................................................................... 121 5 6 Au Calibration curve for concentration range of 100267.5 nm ............................................................................................................................... 122 5 7 Representative Ca spectra 1g/ml ...................................................................................... 123 5 8 Ca calibration curve for concentration range of 0.5 ............................................ 124

PAGE 11

11 5 9 Image of plasma emission showing 532nm probe laser scattering from a single ~50 m Lu -doped droplet particle. .......................................................................... 126 5 10 Representative spectra of M g corresponding to delay times of 5, 15 and 30 .......................................................................................................................................... 127 5 11 Representative spectrum of Lutetium corresponding to delay times of 1,10 ............................................................................................................................... 127 5 12 Boltzmann plots for Lu II lines at 347.3 nm, 350.7 nm and 355.7 nm respectively. Upper energy levels for these lines are 41225, 28503, 45458 cm1 respectively. ................................................................................................................. 129 5 13 Temporal temperature evolution of the analyte by introducing single droplet of Lu at 1 Hz repetition rate. ............................................................................................... 130 5 14 Ion -to -Neutral ratio of Mg lines at 280.27 nm and 285.21 nm respectivel y. Mg is mixed with Ca in ratio of 1:5 in this experiment and overall conc. of 3,000 g/ml. ......................................................................................................................... 131 5 15 Silicon atom temperature plotted against residence time by introducing single droplet of S i analyte in a ICP MS system at 1 Hz repetition rate. ......................... 132 5 16 Spectral images of hydrogen diffusion measured temporally. Hydrogenalpha line filter (656.28 nm) was used with line width of 10 nm. The bar represents 1 mm. .................................................................................................................. 135 5 17 Average diameter of Hydrogen emission burst as a function of delay time. Error bars depict standard deviation, where N= 50. ........................................................ 135 5 18 Spectral images of Calcium diffusion measured temporally. Calcium line filter (396 nm) was used with line width of 3 nm. The bar represents 1 mm. ................. 137 5 19 LIBS response for Ca as a function of delay time ............................................................. 138 6 1 Quantification of plasma initiation processes. ................................................................... 151

PAGE 12

12 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 LASER INDUCED PLASMAS AS AN ANALYTICAL SOURCE FOR QUANTITATIVE ANALYSIS OF GASEOUS AND AEROSOL SYSTEMS: FUNDAME NTALS OF PLASMAPARTICLE INTERACTIONS By Prasoon K Diwakar August 2009 Chair: David W. Hahn Major: Mechanical Engineering Laser induced Breakdown Spectroscopy (LIBS) is a relatively new analytical diagnostic technique which has gained serious attention in recent past due to its simplicity, robustness, and portability and multi -element analysis capabilities. LIBS has been used successfully for analysis of elements in different media including solids, liquids and gases. Since 1963, when the first breakdow n study was reported, to 1983, when the first LIBS experiments were reported, the technique has come a long way, but the majority of fundamental understanding of the processes that occur has taken place in last few years, which has propelled LIBS in the di rection of being a well established analytical technique. This study, which mostly focuses on LIBS involving aerosols, has been able to unravel some of the mysteries and provide knowledge that will be valuable to LIBS community as a whole. LIBS processes c an be broken down to three basic steps, namely, plasma formation, ana l yte introduction, and plasma an a lyte interactions. In this study, these three steps have been investigated in laser induced plasma focusing mainly on the plasma particle interaction s U nderstanding plasma -particle interactions and the fundamental processes involved is important in advancing laser induced breakdown

PAGE 13

13 spectroscopy as a reliable and accurate analytical technique. Critical understanding of plasma particle interactions includes study of the plasma evolution, analyte atomization, and the particle dissociation and diffusion. In this dissertation, temporal and spatial studies have been done to understand the fundamentals of the LIBS processes including the breakdown of gases by the laser pulse, plasma inception mechanisms plasma e volution, analyte introduction and plasma -particle interactions and their influence on LIBS signal. Spectral measurements were performed in a laser induced plasma and the results reveal localized perturbat ions in the plasma properties in the vicinity of the The measurements provide direct evidence of matrix effects in the LIBS plasma at early times. Electron density measurements at very early times in the plasma reveal deviations from the Local Thermodynamic Equilibrium conditions (LTE) The data from various experiments suggests a complex interaction between the plasma and the aerosol particle, during which the finite time -scales of particle dissociation, and heat and mass transfer are fundamental processes.

PAGE 14

14 CHAPTER 1 LIBS Introduction LIBS a future superstar this was part of the title of a review article on various analytical spectrometric techniques by Winefordner et al. which signifies the growing importance of Laser Induced Breakdown Spectroscopy (LIBS) in the analytical spectrometry community [1]. With time, the application of LIBS in elemental analysis has been ever increasing. The history of LIBS has a checkered past. Dating back to 1960s, LIBS developed from a micro-sampling technique to an elemental analysis technique, but sometimes, in between the years, relegating just to a scientific curiosity without much development with regards to the analytical aspects of LIBS [2]. Recently there has been renewed interest in the analytical community in developing this analytical technique. LIBS is an analytical technique which was born with the invention of laser in 1960s. Even though LIBS has been in existence for about 50 years now, prior to 1980, its major application was in the study of plasma formation. Its use for elemental analysis started only later, with the advent of advanced lasers, spectrometers and detectors. In its simplest form LIBS involves focusing a high energy pulsed laser to a very small point, thereby forming plasma in which the resulting atomic emission intensity information is used for qualitative and quantitative analysis. There are various other popular techniques, such as Inductively Coupled Plasma-Mass Spectroscopy (ICP-MS), Inductively Coupled PlasmaAtomic Emission Spectroscopy (ICP-AES), electro thermal atomization-atomic absorption spectroscopy etc., which have been used for various industrial, analytical, biological, environmental applications for elemental analysis. LIBS have been used for liquids, gases, solids, aerosols for toxic metal waste

PAGE 15

15 monitoring [3-5], environment pollutants monitoring[3-6], process monitoring[7], remote sensing[8-9,13-14], cultural heritage applications[10], biomedical applications[11-12], space explorations [13-14]. LIBS provides the advantages of high sensitivity remote-sensing, capability no-sample preparation, robust set up and ability to detect or analyze any element in periodic table. All these advantages along with more understanding of the basic mechanisms and processes can surely make LIBS a superstar or atleast one of the superstars in atomic spectroscopic methods in coming years. LIBS in Different Media: Solids, Liquids, Gases LIBS has been used in all kinds of media with varying success. Typically LIBS was used for solids analysis, initially, but in recent years it has been applied to all kinds of media. In solids, breakdown is initiated by multiphoton ionization or inverse Bremsstrahlung which is followed by cascade ionization, as illustrated in the following processes (discussed in detail in section 1.5): +-++ MehnM )(n (1-1) -+ ehen (1-2) +-++ MeMe 2 (1-3) As compared to solids, breakdown threshold in gases are 2-4 orders of magnitude higher. In gases, less of the energy is required for atomization, so most of the energy is coupled into the excitation process, and hence the ion-to-neutral ratio can be very high depending on laser irradiance. Liquids are usually analyzed by forming plasma on surface or in liquid droplets. Compared to gases, the plasma in liquids decays more quickly, is cooler and emission lines are broadened. Particles trapped in air, also called

PAGE 16

16 aerosols (a suspension of particles in a gas), have been an area of interest especially for environmental monitoring. Aerosol analysis by LIBS provides a new technique for characterization of aerosols but it also comes with many challenges. As with gases, in aerosol analysis a significant part of the laser energy is used for excitation of the sample, leading to brighter spectra, and the possibility of detecting trace elements. Incomplete vaporization has been issue with LIBS in aerosol analysis, as discussed in detail below, along with other related issues like matrix effects and fractionation effects. From here on, the discussion will mainly focus on LIBS for aerosol analysis. The main objective and motivation is to understand the processes and mechanisms taking place when the plasma and particle interact with reference to aerosol analysis and how these understandings can be used to make analysis of aerosol systems reliable and accurate. Challenges in understanding the fundamentals of the plasma processes while analyzing aerosol particles and thereby making LIBS a robust technique for aerosol analysis is the main motivation behind this study. Why LIBS? LIBS like other analytical techniques mentioned above have the advantage of detecting any possible element from periodic table, along with multi-elements analysis. Besides these advantages, LIBS also provides an edge in many applications. Some of them are listed below: No sample preparationFor LIBS no sample preparation or very minimal sample preparation is required, which is very crucial and important in many analyses. Sample preparation takes time and also can alter the composition of sample with time, and thus LIBS is very useful in that regard because this feature also enables online or real time monitoring of sample. Researchers have used LIBS for online detection of ambient air aerosols [6,15], Hydrogen leak detection[8] and toxic metallic waste[3-5] which shows that LIBS can be deployed in the field very easily.

PAGE 17

17 In situ analysisAs discussed above, LIBS is easily deployable and the instrument can be brought to the sample to do the analysis rather than sample being brought to the instrument. Also, since only optical access is needed to the sample, the laser source can be few meters away from sample for analysis and thus detection of hazardous elements is also possible. Simple and robust setup Can be applied to liquids, solids, gases, aerosols Robust plasma which is important to minimize possible matrix effects. Along with the above mentioned advantages of LIBS, it has also certain drawbacks or challenges which need to be addressed in the future to make it a more reliable and robust and accurate technique. Nonhomogeneity of sample and matrix effects are two major issues which need to be addressed. LIBS being a point detection technique, non-homogeneity in the sample can lead to erroneous results as the analyte being sampled might not be representative of the actual concentration. Matrix effects are a potential challenge in making LIBS a quantitative diagnostic technique. Matrix effects arise when the presence of other elements in the sample can change the signal of the element without any change in concentration. Matrix effect can be divided into two typesphysical matrix effect and chemical matrix effect [16]. Physical matrix affect is mainly due to differences in properties like latent heat, vaporization, thermal conductivity etc This can also lead to varying vaporization rates for multi-element sample, and can lead to fractionation effects which will be discussed in detail (in chapter 3). Chemical matrix effect arises when the presence of one element changes emission of the other signal e.g. the presence of highly ioinizable matrix element can alter the signal of other element. This poses serious calibration issues and need to be understood. To date, these issues have not been adequately analyzed for the LIBS technique.

PAGE 18

18 Basics of LIBS Laser Induced Breakdown spectroscopy is an atomic emission spectroscopy technique which involves focusing high-power laser to a very small volume leading to formation of a highly energetic plasma. Temperatures in the resulting plasma can range from 30,000-60,000 K, which atomizes and vaporizes the sample to produce atoms and ions. Every element has signature emission lines which are collected by appropriate optics and dispersed, typically by using spectrograph, and then finally to an appropriate detector, such as ICCD, CCD or PMT. This analyte signal is the LIBS atomic emission which gives information regarding elemental composition of the sample and concentration of the sample. Usually many such signals are ensemble averaged to provide better uniformity in signal. LIBS signals are temporal in nature and it lasts for 10s of 100s of microseconds after the plasma is formed (see Figure 1-2). Appropriate gating and timing of the ICCD camera helps in increasing the signal-to noise ratio by maximizing the signal and minimizing the background continuum emission [17]. Lasers are the most important component of a LIBS system as it dictates the laser-particle energy coupling, and helps to determine the plasma dynamics. A typical LIBS setup is shown in Fig 1.1 which has been used by Hahn and co workers for LIBS analysis of aerosols [18]. Figure 1-1. Typical LIBS experimental setup [18]

PAGE 19

19 Temporal Nature of LIBS Signal As mentioned earlier, LIBS signal is temporal in nature and lasts for bout 10s of 100s of microseconds after the plasmas inception. Initially, continuum emission dominates which is mainly result of Bremsstrahlaung (free-free) and recombination radiation (free-bound) from the plasma as free electrons and ions recombine as the plasma cools down. After the initial bright continuum emission decays, the emission also features atomic emission by ions and then neutrals, and finally by molecules, which is shown in schematic below in Fig 1-2. As seen, the early continuum emission can cause serious issues in Signal-to-Noise ratios (S/N), especially if the element is available as a very trace amount in sample. For that reason, proper gating and integrating the signal is very useful, enabling one to choose the delay time after which continuum emission has subdued and then integrate the signal. Temporal nature of LIBS signal is shown in Fig 1-2. Proper combination of td, the delay time, and tb the signal integration time, provides a good S/N ratio[17]. Laser Pulse Plasma Initiation Ions and Neutrals Continuum Emission Ions and Neutrals Atomic and Continuum Emission Neutrals and Molecules Atomic and Molecular Continuum Emission Recombination Decaying Emission s m ms ns 0 time Laser Pulse Laser Pulse Plasma Initiation Plasma Initiation Ions and Neutrals Continuum Emission Ions and Neutrals Continuum Emission Ions and Neutrals Atomic and Continuum Emission Ions and Neutrals Atomic and Continuum Emission Neutrals and Molecules Atomic and Molecular Continuum Emission Neutrals and Molecules Atomic and Molecular Continuum Emission Recombination Decaying Emission Recombination Decaying Emission s m ms ns 0 time Figure 1-2. Temporal nature of the LIBS signal

PAGE 20

20 Fisher et al. have shown that optimum temporal gating is very crucial and helpful in obtaining for good S/N ratio, and it varies from element to element [17]. They studied six toxic waste elements and found that for elements with atomic emissions at higher upper energy levels (Arsenic, Beryllium, and Cadmium), shorter optimal delay time is required, while elements with atomic emissions at relatively lower upper energy levels (Chromium and Lead) require longer optimal delays for better S/N ratio. The reason being that low upper energy states lead to stable upper energy level populations in the plasma, hence shorter atomic emission decay time, and vice versa. At longer optimal delays, the S/N ratio improves but the signal intensity is diminished and so larger gate widths are employed to counter the effect. In past years various groups have worked to understand and optimize LIBS signal by temporal gating method Laser Induced Plasma Plasma is formed when a high energy laser is focused into a very small volume so that all the laser energy is coupled to the plasma medium, leading to either multiphoton ionization or cascade ionization. Free seed electrons ( e-) are necessary to start the LIBS process which either occur naturally (cosmic rays, natural radioactivity) or as a result of multi-photon ionization (MPI). Multi-photon ionization results from simultaneous absorption of multiple photons by atoms and molecules resulting in ionization, which can be described as +-++ MehnM )(n (1-4) This process requires multiple photons being absorbed by atoms and molecules, thereby releasing an electron, and so no initial population of electrons is required. Also in each process, only one electron is released so the number of electrons does not increase

PAGE 21

21 exponentially as is the case with cascade ionization (discussed below). Since no electron is required to start this process, it is believed that laser induced plasma is initiated by multi-photon ionization, and later cascade ionization follows. Since multiple body collisions exist, probability of multi-photon ionization is low and also requires high energy density. For example, the photon energy of Nd:YAG laser at 1064 nm is ~1.1 ev, while ionization potential for the N2 molecule is 14.5 ev, which implies that for MPI process about 12 Nd:YAG photons will be required, which is an extremely low probability process. In cascade ionization initial free electrons absorb laser energy and collide with neutral atoms and molecules to knockout more electrons. Cascade ionization can be written as: -+ ehen (1-5) +-++ MeMe 2 (1-6) where M represents a neutral atom or molecule. This process is also referred to as avalanche ionization, since the population of free electrons increases exponentially through repeated collisions. In general, both the processes occur simultaneously and lead to initiation and growth of the plasma, but with varying magnitude at different times during plasma life-cycle. Transient Plasma Dynamics Understanding plasma dynamics is the key in understanding many underlying basic mechanisms in the plasma-particle interactions, and can also help in understanding effects like fractionation effects which causes erroneous quantitative measurements. Early plasma dynamics are not clearly understood yet are very challenging to study

PAGE 22

22 owing to the often overwhelming continuum emission, deviations from Local Thermodynamic Equilibrium (LTE), localized non-homogeneities, and so on. Hohreiter et al. have shown that particle dissociation, atomic diffusion, and heat transfer have finite time-scales that result in spatially non-homogeneous (localized) analyte atomic emission [18].Such an effect can cause matrix/fractionation effects at early times in plasma. Through quantitative plasma imaging studies, Hohreiter showed that discrete particles dissociate on a scale of tens of of microseconds after plasma inception, which is comparable to the time scale of spatial non-homogeneity. Also, very early in plasma, high electron density and high electron and ion temperatures can play important role in plasma emission characteristics. High electron density can affect plasma particle interactions and also affect Local Thermodynamic Equilibrium (LTE). In most of LIBS models LTE is assumed wherein electron temperature is equal to ion temperature (Ti=Te). However it is interesting to investigate if there is any deviation from LTE at very early times in plasma, and how it affects emission characteristics. Earlier work by Hahn and co-workers has shown that transient plasma dynamics plays an important role in the vaporization process, and the importance of characterizing the plasma properties [19]. Hohreiter et al. in their earlier work have studied evolution of plasma by calculating electron density, emissivity and spectral absorbity using Stark broadening and transmission experiments during the first few hundreds of nanoseconds [19]. Parriger et al. measured Stark broadened emission profiles of Balmer Series H lines from laser-induced optical breakdown in gaseous hydrogen [20]. Using these measurements, they calculated electron densities from as early as 5 ns delay, where they measured values as high as 1019 cm-3 for earliest times. Villagarn et al. have calculated

PAGE 23

23 electron densities of order of 1018cm-3 and temperature of about 105K during first tens of nanoseconds of plasma formation using 300 mJ energy laser pulse [21]. Because the initial breakdown and growth processes are important in the eventual plasma-particle interactions, it is a good idea to measure and estimate electron density and electron temperature at very early times in plasma. Such measurements may provide additional insight into various processes taking place including diffusion, conductivity and gives a measure of plasma instability. All these processes have direct influence on vaporization of analyte and emission characteristics. Fractionation and Matrix Effect While matrix effects have received considerable attention over the years with regard to LIBS and other analytical techniques, the unique application of LIBS for aerosol analysis has been largely unexplored with regard to this topic. The current study is limited to the implementation of LIBS for analysis of aerosol systems, namely gasphase breakdown in the presence of fine particles. Precise calibration of the analyte spectral signal with respect to the true analyte concentration within a laser-induced plasma is necessary for achieving accurate results with the LIBS technique. A key assumption for LIBS is that complete dissociation of the constituent species within the laser-induced plasma results in independence of the analyte atomic emission signal on the analyte source, as well as independence with respect to the presence of other elements. However, recent work by Hohreiter and Hahn has demonstrated that individual aerosol particles dissociate over a finite time scale of some microseconds, and that finite atomic diffusion rates for particle-derived atoms result in the atoms being localized about the original particle location over these same

PAGE 24

24 time scales [18]. This behavior suggests that aerosol particle-derived analytes may very well experience different plasma conditions (i.e. temperature and free electron density) than the overall average (i.e. bulk) plasma conditions, which may in turn induce matrix conditions not previously considered in the idealized model of analyte independence. This effect of variation of analyte response by different elements within the original bulk material matrix is referred to as fractionation or matrix effects, and has been widely observed and studied in another spectroscopic techniques such as inductively-coupled plasma atomic emission spectroscopy (ICP-AES), and laser ablation inductively-coupled plasma mass spectrometry (LA-ICP-MS). Fractionation or matrix effects render analytical techniques less effective for quantitative analysis. Various studies have been done in the analytical community to study and understand fractionation and matrix effects on the resulting analyte response [22-28]. In an effort to correct for matrix effects, a common analytical procedure involves normalization of analyte signal to a standard reference signal, although appropriate reference signals are not always readily available. For example, aerosol analysis of ambient air presents a widely varying range of particle composition with no invariant species for signal normalization. Alternatively, acoustic signals and temperature normalization coefficients have been used to calibrate the analyte signal as a means of accounting for vaporized mass and temperature excitation variation, respectively, in the plasma [24]. Along these lines, Mermet suggested the use of robust conditions to minimize matrix effects by using the ratio of ionic-to-neutral magnesium atomic emission lines [25].

PAGE 25

25 The LA-ICP-MS technique has become a popular choice for trace elemental analysis of solid samples. With LA-ICP-MS, three sources of fractionation may be considered: during the laser ablation process, during transport to the ICP, and finally within the inductively coupled plasma. Fractionation in the ablation process is attributed to laser-material interaction, and may depend on laser fluence, laser wavelength, and laser pulse width, as well as subsequent analyte transport effects that are related to the resulting size distribution of the aerosol (including nucleation effects), carrier gases, and transport losses due to impaction and diffusion losses [29-35]. Such effects are largely eliminated with direct aerosol sampling via LIBS, as implemented in the current study. More relevant to LIBS analysis is fractionation within the inductively-coupled plasma, which can be attributed to incomplete vaporization, atomization, and ionization processes. These processes are highly dependent on the plasma temperature (both gas temperature and electron temperature) and the free electron density [25, 36-38]. Budic reported the suppression of analyte emission with ICP-AES in the presence of matrix elements, which was attributed to a shift in ionization equilibrium [39]. Gunther et al. have demonstrated that incomplete vaporization of large particles in the ICP (1-1.5 m m) causes elemental fractionation effects in LA-ICP-MS [23]. Significantly, even after optimizing the LA-ICP-MS system for minimization of ablation and transport effects, they found differing analyte responses that were related to the overall particle size entering the ICP, concluding that plasma-particle processes play a major role in analyte response. Detailed study has been done by Olesik investigating the vaporization process and kinetics of droplets in ICP-MS, with a focus on matrix effects [40]. In more recent work, Wang et al. showed that vaporization efficiency and

PAGE 26

26 subsequent radial diffusion are controlling processes that are coupled closely to the plasma gas temperature [22]. While increased plasma temperature can increase particle vaporization, a careful balance must be given to a concomitant increase in atomic diffusion, which can result in an increased analyte fraction missing the MS skimmer. Transient Electron Density Study of electron densities and temperatures in plasma have been an area of study for many decades. Knowledge of electron densities and temperature helps in understanding various instabilities in plasma, plasma equilibrium, and various mechanisms including diffusion, energy transfer and so on. Stark broadening, Langmuir probe, Saha Boltzmann method, interferometer method and spectrometry, are a few of the methods to measure these quantities in plasma [19-21, 41]. All of them have their advantages and disadvantages. Thomson scattering is one of these techniques used to characterize plasma that has been studied for decades, but its use in LIBS community has been limited to date. A basic understanding and review of these techniques are provided here. Stark Broadening Method This technique involves measurement of the full width at half maximum of lines like N and H in the plasma. There are various mechanisms and causes for spectral profile development of different elements, but in typical LIBS conditions, the Stark effect is the main contributor for line broadening. The electric field of electrons influences the energy levels of atoms and ions (Coulomb interaction) and cause it to broaden. Both ions and electrons are responsible for Stark broadening, but owing to higher velocities of electrons, the electrons dominates. This knowledge of line broadening can be used for

PAGE 27

27 electron density measurement. The Balmer series of H lines(82 259.2865 102 823.911 cm-1) is one of the most commonly studied lines for electron density measurement using this method. Electron densities in range 1016-1019 cm-3 have been measured in plasma by using the H and H lines which correspond to temperature range of 7,000 100,000 K [19-21]. Typical H Stark broadened line is shown here in Fig 1-3, which is actually a combination of a multiplet of seven lines. Each line is resolved and fitted with Lorentzian or Voigt profile, to determine the full width half maxima of Stark broadened lines. Figure 1-3. Typical Stark broadening of H line and fitted Voigt profile [42]

PAGE 28

28 Stark broadening of a particular line is given by = 6 3 1 4 1 6 610 1 10 5.32 10e D e en wBN n Aw n b t n f +-, (1-7) where en is electron density, w is electron impact parameter, A is ion broadening parameter, DN is number of particles in Debye sphere, B is a coefficient values ranging from 0.75 -1.2. For LIBS conditions, the ion broadening parameter is negligible, reducing the equations to = w ne2 10 6 and (1-8) en = lD w 2 106 (1-9) The value of w is a function of temperature and has been tabulated by Griem for different elements [43]. LTE is not assumed in this method. In general there are other broadenings, like instrument broadening, and Doppler broadening, which interfere with Stark broadening and makes the measurements less reliable if not accounted for. To interpret the data, Abel inversion maybe required. Abel inversion assumes plasma symmetry, which causes calculation errors, especially in central part of the plasma, which many times limits the use of Stark broadening technique for measurement of electron density. Nonetheless, it remains a very popular method for electron density measurements given the relative simplicity of the technique. Saha-Boltzmann Method In this method, electron density is measured by ratioing the line intensity of two ionization states of same element. LTE is assumed for this method, which implies

PAGE 29

29 confirmation with both the Boltzmann and Saha equations. The electron density expression by this method is given as () ( ) KT EEEion I m II i II mn I ij II ij I mn e eI m II ie g g A A I I h KTm n-+ -=3 2 322p (1-10) where em is the mass of electron i, m excited energy levels, ijA, ig are transitional probability and statistical weights of upper level I, and I is intensity emitted[16]. The Griem criterion can be used to assess whether LTE will exist in plasma for particular atomic and ionic transitions [16]: () 3 2 2 1 317)( )(545,30)10( b t n n f -nm z KT cmne el, (1-11) where en is lower limit of electron density for LTE, z is degree of ionization (its value is z=1 for atoms and z=2 for singly charged ions). While this condition is generally satisfied for LIBS conditions, there are other criteria which may not be fully met for LTE and therefore makes Saha-Boltzmann method prone to errors in T and en calculations. Electron temperature and density are coupled together, so any error in measurement of one propagates the error in measurement of other. Also the assumption Ti=Te (where Ti is ion temperature and Te is electron temperature) is not valid for plasma conditions deviating from LTE conditions. As discussed above, Abel inversion maybe needed to resolve the data spatially, and the inherent symmetry assumption leads to errors in centre of the plasma. Also the transition probability values are not always available leading to problems in the electron temperature and density measurement by this technique.

PAGE 30

30 Langmuir Probe Another technique to measure Te and ne is by using physical probes called Langmuir probes inside the plasma. These probes measure the electron velocity profile, which can then be used to measure fundamental properties of the plasma. This technique is simple and easy to use, and may have better spatial resolution than other optical techniques discussed above. However being invasive in nature ,it disturbs the plasma and can alter the plasma conditions which can give rise to error in measurements. Single Droplet Study Various fundamental processes involved in LIBS can be observed and studied separately by measuring signal intensity, ions and electrons from a single analyte droplet. Using spatial and temporal measurements, kinetic information on parameters like desolvation, vaporization, atomization and ionization, etc. can be deduced. In general, nebulizers produce hundreds and thousands of droplets of varying sizes in plasma volume per second and hence single droplet study is not possible in such an arrangement. Optical trap is one technique which can be used to optically trap a single droplet of analyte and perform LIBS study. Radiation pressure applied by high energy laser is able to trap small particles. Schematic of such an arrangement discussed by Ashkins et al. is shown in Fig 1.4 [44-46]. Even though it is theoretically possible and has been demonstrated in various laboratories, its implementation is extremely tricky and difficult to achieve for smallest of particles (<1 m). Alternative technique can be the use of droplet generator, which can be used to produce single droplets which can be introduced in the plasma. Scientific droplet generator offer good repeatability of droplets at same speed, it can give excellent replication of single droplet experiment as in optical trap. The motivation behind such

PAGE 31

31 kind of study is to improve the reliability of LIBS measurements. Imaging of single droplets in plasmas and analyzing the resulting signal can provide information regarding spatial dependency on analyte signal. Simpson et al. showed spatial dependency of the particle in plasma for maximum signal in their single particle experiment [47]. Figure 1-4. Typical Optical trapping set up. [46] The above discussion makes it clear that plasma-particles interactions are a very complex process involving various fundamental processes including vaporization, atomization, laser-plasma interaction, desolvation etc which govern plasma dynamics and thereby LIBS signal. Especially during early times in plasma, the processes are not well understood. Early plasma dynamics are dominated by high electron density and high electron temperature gradients, LTE deviations, localized effects, spatial effects etc. which influences quantitative measurement. This study aims to find answers to few of the many questions related to plasma-particle interactions and its implications to LIBS.

PAGE 32

32 It is hoped that the results extend beyond the LIS technique to other related atomic emission techniques including LA-ICP-MS, LA-ICP-AES. Motivation and Objectives As discussed earlier, the plasmas form an integral part of many analytical techniques including LIBS, ICP-MS, ICP-AES etc. Plasma-particle interactions and plasma dynamics play an important role in the analyte response which is important to understand with regards to quantitative analysis of the analyte. While LIBS has been shown to be an effective qualitative diagnostic technique, it suffers from various uncertainties, when quantitative measurements are performed. Fractionation and Matrix effects lead to erroneous analyte concentration measurements. These effects can be minimized or eliminated altogether if greater understanding of the fundamental processes taking place during the plasma-particle interactions is achieved. These uncertainties arise because of various assumptions involved in the treatment of the plasma-analyte interactions, notably the assumption of Local Thermodynamic Equilibrium in the plasma. LTE assumption provides simplification in the analysis of plasma dynamics. But the plasma-particle interactions are transient in nature and there can be regions, temporally and spatially, where the assumption of LTE is not valid, and thereby can lead to incorrect interpretations. Study of the plasma-analyte interactions is an intriguing and interesting problem which is a challenge as well as the motivation for understanding the basic physics involved in the process. Decoupling various fundamental processes in the plasma-particle interactions, and resolving them temporally and spatially, to provide better understanding, is the main objective of this study. LIBS, with the advantages mentioned earlier, combined with improved reliability and accuracy can be the next superstar in analytical methods for elemental diagnostics.

PAGE 33

33 This can be achieved only by understanding the plasma interactions in its entirety, which serves as the goal and motivation for this study. Such an understanding will not only benefit LIBS community, but all the other analytical techniques which employ plasma as the source of excitation. With this motivation, the objectives of this study are outlined as follows: Spatial and temporal study of the plasma-particle interactions. Study and investigation of various underlying assumptions in the plasma-particle analysis including LTE, plasma robustness, homogeneity etc. To decouple the fundamental processes in the plasma-analyte interactions including measurements of local excitation temperature and diffusion coefficients by designing single droplet experiments.

PAGE 34

34 CHAPTER 2 AEROSOL SAMPLING POISSON DISTRIBUTION Introduction Sampling of aerosols with the LIBS technique differs from sampling in other analytical techniques. Usually in many analytical techniques, the analyte sample is introduced as a continuous stream into the plasma and the resulting signal is integrated in time, which gives measurement of the average analyte concentration in the plasma volume. Consider for example, Inductively Coupled Plasma-Mass Spectroscopy (ICPMS), where the plasma is a continuous source, and the analyte is introduced in the plasma volume at a continuous rate. Comparing ICP-MS, or for that purpose other analytical techniques, to aerosol sampling with LIBS, reveals a totally different scenario. LIBS plasma volume is finite in nature and coupled with the discrete nature of aerosols, a complex sampling problem emerges and calls for the need of a different approach for signal analysis than the traditional ensemble averaging methods. For aerosols, such sampling and analysis method may potentially offer no response or extremely weak response for the presence of dilute analyte in the sample. In order to increase the response for LIBS analysis, a sequence of signals (i.e. single spectra) is recorded and each LIBS shot is assumed to sample a similar sample volume in plasma. These individual shots can later be averaged or conditional analysis can be performed to provide a measure of analyte concentration. For aerosol analysis, each individual LIBS spectrum may provide additional information regarding the discrete analyte nature and by optimal gating, signal collection, and processing schemes, aerosols can be studied using LIBS very effectively.

PAGE 35

35 Conditional analysis of aerosols provides the opportunity to discard signals which have no analyte response while only those spectra are considered for analysis which have an analyte response beyond a threshold limit. This technique has been developed and discussed in detail in earlier works of Hahn and co-workers [6, 48]. By using this technique with appropriate algorithms, single shot LIBS for aerosol analysis can be possible. Discreteness of the analyte in the plasma volume is very important when considering LIBS for aerosols, for example when sampling exhaust plumes containing metals since most of the metals in these wastes are in particulate form and thus can be sampled discretely. The exception of mercury with its high vapor pressure. In this study, goals are to measure dilute aerosol particles with regard to understanding the sampling process and sampling statistics. Ambient air serves as a nice aerosol source for such a study. Average number of aerosol particles in the sampling plasma volume is given by = N Vplasma (2-1) where N is the number density of aerosols (particles/volume of gas) and Vplasma is the effective plasma volume, which for current experiment has been calibrated and found to be of order of 1 mm3[49].The number density of aerosols is related to mass concentration, C (aerosol mass/volume of gas), of aerosols by following relation 3 6 vmd CrN p r= (2-2) where is the bulk density of aerosol particles, rvmd is the mean volume diameter of the particle, which is calculated by integrating over the normalized particle size distribution function p(r) as given by,

PAGE 36

36 3/1 0 3)(r = =r vinddrrprr (2-3) Combining the above two equations, one can relate the average number of aerosol particles to concentration of the sample as 36vind plasmar CVpr m= (2-4) Discrete aerosol sampling in LIBS can be modeled as Poisson distribution, which is given as mm-=e n Pn n (2-5) where Pn is the probability of sampling n number of discrete aerosol particles in a given plasma volume. It follows that the sampling rate, R, is defined as the probability that the laser plasma hits at least one particle in the plasma volume, which can be readily obtained by calculating the Poisson probability for hitting zero particle and then subtracting it from 1 and by summing over all sampling events as R = (1-P0) = = 1 n nP (2-6) This modeling of aerosol sampling in LIBS plasmas using Poisson statistics enables contributions of aerosol as discrete particles in the context of single shot LIBS analysis. This enables one to look into key regimes related to relevant concentrations and particle sizes, and thus identify appropriate algorithms like ensemble averaging or conditional analysis, or some other techniques which can be applied. As has been shown in earlier studies, sampling rates can get as low as 1% and even well below, when studying ambient air [48]. Clearly for such low sample rates, ensemble averaging will

PAGE 37

37 not result in any useful signal and this call for the need of conditional analysis algorithms. The essence of this discussion is that the analyte sampling rate is determined by aerosol concentration and size distribution, and which can be very low for many realistic cases like ambient air sampling. For such cases it is important that LIBS aerosol analysis be analyzed and modeled taking into consideration discrete particles and discrete plasma volume. The discreteness of aerosol particles may provide sufficient information in single LIBS shot even when the sampling rate is extremely low. Hahn et al. have shown that, conditional analysis with proper algorithm can be utilized to analyze such low sampling rate problems[48].With conditional analysis, LIBS shots without any signal are rejected and only those shots with hits are counted and averaged. Suitable conditional analysis algorithm involves determining the threshold to determine the LIBS signal which can be labeled as a hit and secondly to determine the sufficient number of hits so that the signal is representative of aerosol sample. It has been shown that a minimum of 20 shots gives a good representation of a typical aerosol sample using Monte Carlo simulations [50]. Statistical considerations are important in LIBS sampling with regards to aerosol analysis, although to date no systematic measurements have directly assessed the applicability of Poisson distribution. Such an analysis is one goal of the present work. Poisson Distribution As discussed earlier, aerosol sampling in LIBS may be modeled using Poisson distribution, which is defined as a discrete probability distribution used for modeling of random samples or events which are rare and independent from each other [49]. The number of events during any given interval is given by the Poisson distribution

PAGE 38

38 mm-=e n Pn n (2-7) where is a positive real number and gives the expected number of occurrences of the rare event during a given interval. also gives the variance of the distribution which implies that the mean number of occurrences fluctuate about the mean value by standard deviation m, which are also referred to as Poisson noise. The value of single occurrences sometimes might be too small to measure, and in such cases correlation of Poisson noise with the mean value gives a measure of contribution of single occurrence to the whole process. The features of the Poisson distribution can be useful in analyzing and modeling single shot LIBS, and can helpful in identifying key regimes in aerosol sampling in LIBS. Sahara Dust Storm Experiment On July of 25th-26th of 2005, Southern Florida was hit by massive dust storm which originated from the Sahara desert in Africa. Usually about 10-12 dust storms travel across the Atlantic every year, but this one was more prominent and its effect was seen in both southern and central parts of Florida as the sky was characterized by haze and more reddish sunsets. This dust cloud, which originated in the Sahara, had a length scale of about 300 miles, and it took about 10 days to travel to Florida. There were some concerns about these storms causing respiratory problems because of the increase in particulate matter in atmosphere. Also it has been studied that these dust storm adversely affect coral reefs in Florida and Caribbean. Figure 2.1 shows dust concentration in various years because of these storms and how it has affected the coral reef ecology [51].

PAGE 39

39 In order to study the sensitivity of the LIBS instrument, experiments were setup to sample ambient air and see if the LIBS instrument was able to monitor or detect any increase in signal due to particulates in the ambient air. Experiments were planned so as to sample the air a day prior to the onset of the storm, and then a day after the storm. As the storm moved closer to Florida, it also drifted further towards the south, and hence diminished the impact in the Gainesville area. Figure 2-1. Effect of Dust concentration on coral ecology in Florida and Caribbean [51] Experimental Setup The experimental setup similar to our previous work was used here, which is shown schematically in Fig 2.2 The LIBS setup consists of a Q-switched Nd:YAG laser operating at the fundamental wavelength of 1064 nm (5 Hz repetition rate), 10 ns pulse width, laser energy of 290 mJ/pulse, which was focused to create the plasma using a 50mm diameter,75mm focal length lens on to the centre of six-way cross LIBS chamber. The LIBS signal was collected on the axis of the incident beam at an angle of 1800 (backscattering) using a pierced mirror and 75mm condensing lens, and then

PAGE 40

40 focused to a fiber optic and then coupled to a 0.275-m spectrometer (2400 groove/mm grating, 0.15-nm resolution). Backscattering mode of signal collection ensured minimal spatial variation in LIBS signal. Finally the plasma emission was recorded on an ICCD detection array, which was synchronized to the laser Q-switch, and further connected to the computer for analysis. Figure 2-2. Experiment setup for ambient air study [6] As shown in Figure 2.2, ambient air was drawn into the chamber through US EPA PM 2.5 sampling inlet(Rupprecht and Patashnick), which only lets in particulate sizes of 2.5 microns or less.This particular sampling inlet and location has been described in detail by Hettinger et. al. The inlet, with a constant sampling rate of 1 m3/h, was located 5 m above the ground adjacent to a three story building. The transfer line, as shown in

PAGE 41

41 the schematic, was used to transfer the sampled air from inlet to the LIBS chamber ensuring minimal transport losses. The outlet of six-way cross chamber was connected to vacuum pump which provided bulk sampling flow rate from the inlet to the LIBS. Pressure in the chamber was maintained at 1 cm of water below atmospheric pressure [15]. Ionic Calcium emission lines at 393.37 nm (0-25 414 cm-1) and 396.85 nm (0-25 192 cm-1) were chosen to analyze the dust data with fixed detector delay of 40 s and integration time of 40 s. This provided the optimal Signal-to-Noise ratio for Ca lines owing to decay of continuum emission, as has been discussed in various studies [15,17,18]. The dust storm was expected to arrive at the Florida coast around 26-28 of July, and was supposed to stay for about 10 hrs duration in the vicinity of the Florida coast. Sampling of ambient air was started from 25th July evening to provide a baseline before the onset of dust storm. Data was taken in three sessions during the day (morning, noon and evening), all ranging between 9 AM 6 PM. Three different sessions were chosen so as to study the influence of dew, humidity and temperature on the LIBS signal, as dew content during the mornings is expected to decrease the concentration due to settling of the dust. During each LIBS session, 20 sequences data collection was performed consecutively, each sequence involving 1000 laser shots at frequency of 5 Hz. Each 20 sequence lasted for about 3:20 min. The total experiment run time for one data-set was about an hour and corresponds to 20,000 total laser shots (20x1000).The expected particle sampling rate of ambient air by LIBS is expected to be 1% or even lower, which required conditional analysis algorithms to overcome the low signal-tonoise ratio. Conditional analysis was used to reject the spectra without any significant

PAGE 42

42 signal from the analyte. To eliminate any false hits, a threshold was determined for the signal without any analyte present. To calculate the threshold value for the rejection of false hits, the Peak/Base was calculated at the spectral wavelength of interest (Ca lines), when no analyte was present in the chamber (i.e purified air only). Based on this Peak/Base value, then threshold was set and continuously increased until there were zero hits when no analyte was flown. For this study, the threshold was set to be about 100% of the P/B in absence of any analyte in the chamber. This ensured the system to record only real hits, as more than 95% of false hits were rejected [6]. This algorithm helped to ensure that only true hits were recorded while minimizing the loss of data as summarized, by Iida et al [52]. The spectra which correspond to real hits were recorded, and finally the ensemble average was taken and using a calibration curve the equivalent concentration of the analyte sample was calculated. In order to calculate the actual mass concentration, the equivalent concentration was multiplied with frequency of hits (number of real hits/ number of laser shots) Results and Discussions Calcium lines at at 393.37 nm and 396.85 nm were chosen to analyze the dust data. The reason this particular line is chosen because firstly, dust is mainly comprised of calcium particles, and secondly calcium lines are very intense, so that even in low concentration considerable signal can be obtained. For example, recent work with aerosolized spores revealed a calcium detection limit of about 2 femtograms. This will correspond to a single calcium carbonate particle of size ~200 nm, where the Ca containing aerosol particle has been modeled as CaCO3 a common mineral in the ambient, with a density of 2.7 gm/cm3 and Ca mass fraction as 0.40 [15]. Fig 2.3 shows raw spectral data obtained by ensemble averaging 20,000 shots, and by conditional

PAGE 43

43 analysis of 37 real hit shots based on the detection of calcium atoms emission. Number density can be estimated by considering the plasma volume (1.2 mm3 as per ref 49) and LIBS based sampling rate given by ratio of the real hits and number of laser shots 0 2000 4000 6000 8000 1 1041.2 104385 390 395 400 405 410Intensity (a.u.)Wavelength (nm) Ca Ca Avg of real hits only Ensemble average of all hits Figure 2-3. LIBs spectra recorded in ambient air (37/20000), which gives an average number density of 1.54 particles/cm3. It clearly shows that conditional analysis algorithm increase the signal-to-noise ratio, which is consistent with earlier works by Hahn et al., where they were able to measure multiple species in ambient air [6,15] Fig 2.4 shows calcium concentration calculated at different times during the day, starting from 25th of Jul evening to 29th July noon. There is slight increase in concentration of calcium on 27th of July, though it is not significant enough to attribute it conclusively to the dust storm. It is also observed that the mornings of each day had a

PAGE 44

44 higher concentration as compared to noon and evenings, and can be attributed to moisture and dew content in the night air leading to settling of dust closer to the surface of earth. As already mentioned, the dust storm, while reaching Florida coast, drifted far south and hence the LIBS signal in Gainesville did not reveal any significant in change concentration of ambient air dust particles. The maximum concentration obtained was 0.25 g/m3(ppb) which occurred on 27th Jul morning. 0 0.05 0.1 0.15 0.2 0.25 0.3 25E26M26N26E27M27N27E28M28N28E29M29NConcentration (ppb)Dust sampling period from 2529 July Figure 2-4. Dust concentration during 25th Jul-29th Jul 2005 Even though Ca concentration data was not that revealing regarding the dust storm, the data presents an interesting opportunity to examine the applicability of the Poisson distribution. For each session, LIBS data was collected in 1000-pulse laser sequences, for a total of 20,000 laser pulses (i.e. 20 set of 1000 shots each, as described

PAGE 45

45 above). The particular data set which is being discussed here corresponds to the noonsession of 26th July, 2005.Since during this period, 37 spectra were identified as containing a pronounced calcium atomic emission peak; hence they were considered to represent the sampling of a calcium rich aerosol particle. Assuming a plasma volume of 1.2 mm3 [49], the corresponding number of sampled particles and total volume of air sampled by the plasma (1.2 mm3/pulse x 20,000 pulses) yields the measured number density of calcium rich aerosol particles. This value is calculated as 1541 calcium-rich particles/liter of air. The Poisson parameter is readily calculated as = 0.00185. This value of may then be multiplied by the number of laser pulses per sampling interval, namely 1000, to yield the expected value of 1.85 particle hits per 1000 pulses. This value of 1.85 may then be used to predict the sampling distribution of calciumbased particle hits per 1000 pulses. The experimental and theoretical values are tabulated in Table 2-1. Table 2-1. Experimental and Poisson sampling probabilities Hits/1000 Events No. of real hits Exp probability (%) Poisson probability (%) 0 2 0 10 15.72 1 7 7 35 29.09 2 5 10 25 26.90 3 4 12 20 16.59 4 2 8 10 7.67 5 0 0 0 2.84 6 0 0 0 0.88 7 0 0 0 0.23 8 0 0 0 0.054 Total 20 37 100 99.99 This distribution is shown in Fig 2.5, along with experimentally measured sampling distribution over the twenty 1000-pulse data collection intervals. The ideal

PAGE 46

46 Poisson distribution and the experimental sampling rates are in excellent agreement. Specifically, the probability of recording zero particle hits is equal to 15.7 % and 10 % for the experimental data and Poisson distribution fit, respectively, while the most probable sampling rate of 2 particles per sequence corresponds to 29.1 % and 35% for the predicted and experimental data, respectively. Overall the plot demonstrates the statistical nature of LIBS-based aerosol sampling and provides corroboration of Poissonbased models to describe the sampling problem. Figure 2-5. Sampling probability of aerosol in LIBS The aerosol particle sampling rate enables an examination of the LIBS based aerosol analysis problem in the context of discrete aerosol particles, and a finite number of discrete plasma sample volumes. This study is useful for elucidating key regimes suited to ensemble averaging, and those suited to more sophisticated data analysis approaches due to aerosol sampling limitations. As shown above, aerosol analysis with LIBS as modeled by Poisson distribution, may suffer from low sampling rates. Ensemble averaging of analyte response from such low sampling rates lead to a drastic decrease in sensitivity of the technique. Alternative analysis methods and strategies need

PAGE 47

47 to be developed, opposite to the traditional ensemble averaging approach with regards to aerosol analysis with LIBS. Already in this chapter, conditional analysis has been introduced with regards to maximization of LIBS signal from aerosols. Fig 2.5 corresponds to a sampling rate when 37 particles were hit during a sequence of 20,000 laser shots, implying about one particle was hit per 550 laser shots. This means that most of the spectra results in null shots where no analyte particle was hit and the signal therefore contains no information regarding the analyte source. Fig 2.6 shows single shot LIBS spectra, where analyte was sampled, compared to the ensemble average of 20,000 shots on the same scale. It is readily apparent that, ensemble averaging is dominated by null shots leading to loss of valuable information contained in single shot spectra from analyte rich aerosol particles, again emphasizing the need for conditional analysis approach. Figure 2-6. Comparison of single shot spectra to ensemble average of 20,000 shots

PAGE 48

48 To conclude, an appropriate conditional analysis method needs to address the following fundamental issues: Threshold criteria: proper choice of threshold criteria is important, so that all the blank or null shots are rejected, without losing real signal in the process. Hahn et al. have developed an elaborate threshold determining criteria for such conditional analysis approach [6,15, 48]. Generally, Peak-to-Base ratio or Signal-to-Noise ratio has been used for determining the threshold criteria, where the threshold value is determined by increasing the threshold value (usually to more than 100% of the P/B or SNR at the region of interest when no analyte is present) until no false hits or an acceptable numbers are recorded in the absence of the analyte. Statistical considerations: It is important to devise the conditional analysis method keeping in the mind that the number of particles sampled be such that the valid statistical sample of the aerosol is achieved. This ensures that the signal recorded truly represents the actual analyte sample distribution. Hahn et al. have shown that 20 or more number of particle hits ensures a such statistical considerations [49]. Given the discrete nature of aerosol particles and the LIBS plasma, Poisson statistics provide an accurate means to model and predict sampling rates (both hits and null data rate) from expected aerosol number density and the actual LIBS plasma volume. Such an approach is very useful for a priori estimation of expected sample rates thereby giving insight into the use of conditional or ensemble averaging schemes. Calibration considerations: In order to process the resulting spectrum obtained from such conditional analysis methods, appropriate means should be identified so that the spectrum can be used in accordance with calibration schemes used n traditional LIBS, to provide meaningful results.

PAGE 49

49 CHAPTER 3 FRACTIONATION AND MATRIX EFFECT Introduction Various studies have been done in the analytical community to study and understand fractionation and matrix effects on the resulting analyte response, which have been discussed in detail in Chapter 1 [22-28]. As discussed in the introduction (Chapter 1), the roles of particle vaporization, dissociation, ionization, and diffusion are common to a range of analytical techniques, notably the direct analysis of aerosol particles with LIBS and LA-ICP-MS. Previous studies have demonstrated the laserinduced plasma robustness to aerosol sampling, including breakdown initiation and plasma temperature and electron density [53, 54]. However, our more recent work has shown that plasma-particle interactions are confined to localized regions about the particles, thereby clarifying the differences between bulk plasma properties (which have been shown to be independent of the presence of aerosol particles) and plasma conditions in the immediate vicinity of individual aerosol particles [18]. Olesik showed that localized, incomplete desolvation of aerosol droplets affected average time integrated emission intensity [40]. With this framework in mind, it becomes apparent that small amounts of analyte contained in submicron and micron-sized particles can bring about changes in localized plasma temperature, electron density, and ionization fractions, thereby affecting the analyte response. It is the goal of the present study to further understand how particles may perturb the local plasma conditions, and how such perturbations may result in a particle matrix dependent analyte response with LIBSbased aerosol analysis. Results and discussions in this chapter follow from the published work of Diwakar et al [57].

PAGE 50

50 Experimental Methods Figure 3-1. Top view of experimental apparatus for the LIBS aerosol experiments [57]. The experimental system for the current study is shown schematically in Figure 3.1 For all experiments, a Q-switched Nd:YAG laser operating at the fundamental wavelength (2-5 Hz repetition rate and 260 mJ/pulse) was used to create the plasma using a 50-mm diameter, 75-mm focal length lens. The plasma emission was collected on axis with the incident laser beam using a pierced mirror and 75-mm focal length condensing lens. The plasma emission was then fiber-coupled to a 0.275-m spectrometer (2400 groove/mm grating, 0.15-nm resolution). Spectral data were recorded using an intensified CCD detector array. For all experiments, the ICCD was synchronized to the laser Q-switch, and a series of detector delay and integration times were used: 2 m s delay with a 0.4 m s width; 5 m s delay with a 1 m s width; 15 m s delay with a 3 m s width; 25 m s delay with a 5 m s width; 30 m s delay with a 5 m s width; 40 m s delay with a 8 m s width;

PAGE 51

51 and 60 m s delay with a 12 m s width. For each experimental condition, 1000 laser shots were averaged together to produce a representative spectrum, and the measurements were repeated from 5 to 8 times over different days. Aerosol Generation All analyte samples flowed through a standard six-way vacuum cross at atmospheric pressure, which functioned as the LIBS sample chamber as previously described [55,56] and shown in Fig 3.2 Figure 3-2. Aerosol generation system [55,56] A gaseous co-flow of 44 lpm of purified air was used for all experiments. The air was passed through an activated alumina dryer, a course particle filter, an additional desiccant dryer, and finally a HEPA filter cartridge prior to entering the sample chamber. All flow rates were controlled with digital mass flow controllers. The aerosol

PAGE 52

52 particles were made by nebulizing a solution of the desired analyte at a rate of about 0.15 ml/min using a gas flow of 5 lpm through a pneumatic type nebulizer (Hudson model #1724). All analyte solutions were prepared by diluting ICP-grade analytical standards of 10,000 m g analyte/ml (SPEX CertiPrep) to the desired concentration using ultrapurified deionized water. The solution concentrations were adjusted to provide a nominal range from about 1,500 to 25,000 g analyte/m3 through the LIBS sample chamber. Based on previous TEM measurements using the current configuration [55], the average aerosol particle size following droplet desolvation (i.e. solid analyte phase) is expected to be less than 100 nm, while agglomerate formation is considered insignificant. For the current study, analyte species included cadmium, magnesium, and sodium. To verify size and composition for these elements, particles were sampled from the LIBS chamber directly onto carbon-coated 150-mesh copper TEM grids (Electron Microscopy Sciences, FCF150-CU50) and analyzed using TEM and EDS. TEM image of Cd: Zn solution is shown in Fig 3.3. The image shows Cd: Zn particles which were generated by nebulizing and drying solution of ratio 1:4 by mass. EDS measurements revealed the presence of sub micron-sized aerosol particles as binary mixture of Cadmium and Zinc. Overall, the system provided a dispersed, submicron-sized analyterich aerosol stream for LIBS analysis. Results and Discussion Analyte Emission Enhancement with Added Aerosol Mass Results and discussion in this section follows from the published results by Diwakar et al [57].The first set of experiments was performed to assess the change in analyte emission response from an aerosol-derived analyte with the addition of

PAGE 53

53 concomitant mass to the aerosol particles. Specifically, measurements were performed using either sodium or magnesium as the analyte, with copper, zinc or tungsten Figure 3-3. TEM image showing Cd: Zn particles generated by nebulizing and drying 1:4 mass ratio solution. used as the concomitant element. Copper, zinc and tungsten were selected due to their range in volatility, noting their corresponding temperatures for a vapor pressure of 10 torr are equal to 1870, 590 and 4490 K, respectively, as compared to 546 K for sodium. For the sodium experiments, solutions of 1,000 m g Na/ml were nebulized and the atomic emission of the sodium doublet at 589.00 and 589.59 nm doublet (0 16973 and 0 16956 cm-1) was recorded. The spectral signal was quantified as the peak/base ratio

PAGE 54

54 (P/B), where the peak was the integrated full-width intensity normalized by the continuum emission intensity, as interpolated using either side of the sodium peaks. The continuum emission near the sodium lines was featureless and rather flat, making peak integration and continuum interpolation straightforward. The experiments were then repeated for solutions of 1,000 m g Na/ml with an additional 9,000 m g/ml of either Cu, Zn, or W, which corresponds to a Na:Cu, Na:Zn, and Na:W mass ratio of 1:9. For the magnesium experiments, solutions of 500 m g Mg/ml were nebulized and the atomic emission of the magnesium ion at 280.27 nm (0 35,669 cm-1) and the magnesium neutral at 285.21 nm (0 35,051 cm-1) were recorded. As described above with the cadmium spectra, the P/B ratios of the magnesium lines were calculated. The magnesium experiments were then repeated for solutions of 500 m g Mg/ml with an additional 9,500 m g/ml of either Zn or W, which corresponds to a Mg:Zn or Mg:W mass ratio of 1:19. For these experiments, TEM and EDS analysis was performed to verify the mean particle size less than 100 nm, and the binary composition of the resulting particles, although detailed quantitative analysis was not performed. Figure 3-4 presents the spectrum corresponding to the pure sodium aerosol particles, along with the spectrum corresponding to the multi-component sodium/copper aerosol particles at the 1:9 mass ratio. Both spectra were recorded at a delay time of 15 m s following plasma initiation. Several interesting features are noted with respect to Figure 3.4, namely, that the sodium emission signal is noticeably greater in the sodiumcopper aerosol spectrum as compared to the pure sodium aerosol spectrum, despite the constant sodium mass, and that the plasma continuum emission signal is essentially identical for both. One possible reason for the enhancement in Na line intensities can be

PAGE 55

55 attributed to the fact that Na is easily ionized, notably at early stages of plasma development, and therefore the addition of concomitant species may cool down the local plasma volume in the vicinity of the particle, leading to an increase in the Na I-to-Na II ratio. The nearly identical plasma continuum emission intensity was observed for all time delays, and is consistent with previous investigations in which the plasma Figure 3-4.Sodium emission spectra recorded for pure sodium-based aerosols and for sodium-copper containing aerosols at a 1:9 mass ratio. Spectra were recorded at a time delay of 15 s, and are both presented with the same intensity scale [57]. continuum emission intensity was found to be quite independent of the presence or nature of aerosol particles [53,56,58]. However, the enhancement in emission of the Na I

PAGE 56

56 doublet with additional mass added to the sodium aerosol particles was found to have a temporal dependency. To explore such behavior, the measurements were repeated over the delay times from 2 to 60 m s following plasma initiation. To quantify these Figure 3-5. Enhancement of the sodium doublet emission intensity as a function of delay time with respect to plasma initiation for sodium-copper, sodium-zinc, and sodium-tungsten containing aerosols at a 1:9 mass ratio. Data are normalized to the sodium doublet emission intensity for the pure sodium-based aerosols at each respective delay time. Representative error bars correspond to the standard deviation [57]. measurements, the sodium signal from the two-component aerosols (either Cu, Zn or W) was normalized to the sodium signal from the pure sodium experiments at each respective delay time. The normalized sodium signal then represents the sodium

PAGE 57

57 emission enhancement factor as a function of plasma decay time, with the resulting data presented in Figure 3.5. Figure 3-6. Sodium atomic emission signal in the presence of added copper normalized to sodium-only emission. Sodium-to-metal ratio is varied from 1:1 to 1:9 (by mass) To assess the degree to which the mass ratio of sodium to the concomitant element was important, the sodium/copper measurements were repeated for Na:Cu mass ratios of 1:1, 1:3, and 1:6. Interestingly, the sodium enhancement was rather consistent(as shown in Fig 3-6), with the maximum enhancement factor remaining within 10% of the enhancement factor of 1.53 recorded for the 1:9 mass ratio at a delay of 15 m s, although a slight trend was seen of decreasing enhancement with decreasing mass ratio. Nearly identical behavior was observed with the magnesium atomic emission signals when comparing the pure magnesium-based aerosol signal with the magnesiumzinc and magnesium-tungsten aerosol signal. The emission enhancement was observed

PAGE 58

58 with both the Mg I and Mg II emission lines. The data for the 285-nm Mg I line are presented in Figure 3.7 for the Mg:Zn and Mg:W mass ratio of 1:19. As observed in the figure, the magnesium atomic emission signal was found to be enhanced by the addition of added Zn or W mass, although there was also a significant temporal dependence to this trend similar to the results observed with the sodium aerosol measurements. The data presented in Figures 3.5 and 3.7 clearly demonstrate an aerosol matrix effect, which is the first time a direct effect of an additional aerosol particle constituent (i.e. Cu, Zn, or W) on an aerosol-derived analyte signal has been documented for LIBS-based aerosol analysis. Figure 3-7. Enhancement of the magnesium neutral emission intensity (285.2 nm) as a function of delay time with respect to plasma initiation for magnesium-zinc, and magnesium-tungsten containing aerosols at a 1:19 mass ratio. Data are normalized to the magnesium emission intensity for the pure magnesiumbased aerosols at each respective delay time. Representative error bars correspond to the standard deviation [57].

PAGE 59

59 Radziemski et al. were among the first to perform detailed LIBS-based aerosol measurements using a system with particles sizes estimated to be in the submicron size range [59]. Calibration curves were generated for three analytes, namely cadmium, lead, and zinc, and were characterized by initial linearity followed by various degrees of saturation at higher concentrations, which was attributed to incomplete vaporization of particles. An important finding was the general agreement (within 10%) of lead atomic emission signals of comparable atomic lead concentrations when nebulizing either lead acetate, lead chloride, or lead nitrate. Cadmium revealed a 27% difference in analyte response when comparing nebulized solutions of cadmium nitrate and cadmium chloride Their study did not explore the effect of the mass of concomitant aerosol elements, therefore, one may not attribute their findings to a particular effect. In more recent work, Hohreiter and Hahn found significant differences between the carbon emission signal when comparing gaseous and particulate carbon sources [53]. The differences in analyte response were attributed to the preferential concentration of particulatephase analyte within the rarefaction following plasma expansion, which was further supported by additional double-pulse laser experiments [60]. In another recent paper, Mukherjee et al. proposed an internal calibration scheme for quantitative analysis of nanoaerosols that was designed to eliminate effects due to differing plasma conditions at various analytedependent delay times [61]. Their approach, however, utilizes emission lines arising from the bulk plasma gases which, as discussed below, may not reflect the local environment about the aerosol particle-derived analyte. The current findings, in combination with these previous studies, suggests that fundamental changes in the analyte emission are resulting from changes in the aerosol elemental mass composition.

PAGE 60

60 Importantly, these changes in analyte emission have a strong temporal dependency, with the general trend being a convergence to matrix independence with increased time following plasma initiation. For nearly all measurements reported here, the analyte enhancement factors approached unity for delays times of 60 m s, as seen in Figure 3.5 and 3.7, with the exception being Na emission in the presence of W, which showed an actual decrease at the longest delay time. This overall behavior may be explained in the context of recent imaging studies of single particles within laser-induced plasmas that was discussed above [18]. The fact that aerosol-derived atoms were found to be highly localized about the aerosol particle over time-scales of microseconds suggest that the analyte emission process is not governed by the volume-integrated plasma properties (i.e. bulk properties), but rather the localized plasma environment about the aerosol particle. As these atoms diffuse to a length-scale approaching the overall plasma volume, which is on a time scale of tens of microseconds, the effective plasma parameters of the aerosol-derived atoms must converge to the overall plasma parameters, at which point independence of analyte response on other concomitant species should be achieved. The model suggests that the local plasma conditions about an individual aerosol particle and the bulk plasma conditions will differ; hence excitation temperatures and electron density, and consequently ionization fractions, will lead to perturbations in the analyte emission response. With such conditions localized to the aerosol particles, changes in aerosol mass composition can affect the degree of local plasma perturbation, giving rise to the type of matrix effect observed in the present study. To gain additional

PAGE 61

61 insight, more detailed spectroscopic measurements were performed using a selected magnesium-cadmium aerosol system. Localized Plasma Measurements for Mg/Cd Aerosol Particles Experiments were performed for solutions of 500 m g Mg/ml with an additional amount of cadmium at concentrations of 2,500, 5,500 and 8,500 m g Cd/ml, corresponding to Mg:Cd mass ratios of 1:5, 1:11 and 1:17. The magnesium atomic emission intensity of both the ion (280.27 nm) and neutral (285.21 nm ) lines were i ntegrated. The magnesium ion peak at 280.27 nm was found to overlap with some additional spectral features (see Figure 3.8) at certain delay times. Therefore, to more accurately calculate the integrated peak, the Mg emission profiles were fit to Gaussian functions (Doppler broadening), which were then used for integration. Both Lorentzian and Gaussian functions were assessed, and the Gaussian was found to produce an excellent fit. In addition, the atomic emission of the cadmium ion at 226.50 nm (0 44,136 cm-1) and the cadmium neutral at 228.80 nm (0 43,692 cm-1) were recorded and quantified using the full-width integrated emission peak. These measurements were performed over temporal delays from 2 to 40 m s, which provided emission signals for both the magnesium and cadmium ion and neutral lines with good signal-to-noise ratios. Beyond 40 m s, the intensity of the cadmium ion emission line becomes too weak for quantitative analysis. Particles were collected directly on TEM grids within the LIBS sample chamber for the Mg:Cd mass ratios of 1:5 and 1:17. The collected particles were analyzed with TEM and energy dispersive spectroscopy (EDS). No quantitative image analysis was performed, but the particle size fell primarily within the 50 to 100 nm range, and

PAGE 62

62 importantly, EDS revealed elemental compositions of Mg and Cd in ratios consistent with the original solution mass ratios. The TEM analysis reveals that submicron-sized aerosol particles were produced that are composed of binary mixtures of magnesium and cadmium. Figure 3-8. Magnesium emission spectra recorded for magnesium-cadmium aerosol particles at a 1:17 mass ratio as a function of delay time. All three spectra are presented with the same intensity scale [57]. An advantage of using magnesium and cadmium as analytes is the ability to observe both ion and neutral emission lines in the same spectral window. As observed in Figure 3.8, the ion-to-neutral emission ratio decreases with time as the local plasma temperature and electron density change with increased plasma lifetime. However, because the ionization fraction and the overall emission intensity are coupled via Boltzmann and Saha relations, it is not possible to uniquely determine such properties

PAGE 63

63 from the magnesium spectral data alone. Uncoupling temperature, electron density, and ionization fraction effects is particularly difficult given the finite timescale for particle dissociation, which means that the local (i.e. centered about the aerosol particles) plasma conditions are rapidly changing as energy is absorbed by particle vaporization and ionization requirements, and additional electrons are released. To quantify such perturbations to the plasma conditions, additional spectroscopic measurements were made in the vicinity of the cadmium ion and neutral emission lines. Figure 3.9 shows the spectra recorded in the vicinity of the cadmium emission lines for the three different Mg:Cd mass ratios at a fixed delay time of 25 m s. As observed in the figure, the cadmium emission signal does increase with increasing cadmium mass fraction within the aerosol particles, and as noted above the continuum emission intensity remains identical, again demonstrating the independence of the bulk Figure 3-9. Cadmium emission spectra recorded for magnesium-cadmium aerosol particles at a fixed delay of 25 s as a function of Mg:Cd mass ratio. All three spectra are presented with the same intensity scale [57].

PAGE 64

64 plasma emission on the presence of aerosol. To quantify the emission response of the cadmium aerosol fraction, the cadmium ion-to-neutral (226.5 to 228.8) emission ratios were calculated and are presented in Figure 3.10. The cadmium ion-to-neutral ratio is observed to decrease with delay time, and significantly, the data reveal a trend of increasing rate of decay of the ion-to-neutral ratio with decreasing Mg:Cd mass ratio. Fitting an exponential decay to the three curves, the decay constants are 0.062, 0.050, 0.044 m s-1 for Mg:Cd ratios of 1:5, 1:11,and 1:17, respectively. This behavior is evidence of changes in the local plasma conditions surrounding the Mg-Cd particles, although quantification of the exact nature of such changes will require a combination of the magnesium and cadmium emission data. Figure 3-10. Cadmium ion-to-neutral (226.5-to-228.8 nm) intensity ratios as a function of delay time with respect to plasma initiation for magnesium-cadmium mass ratios of 1:5, 1:11, and 1:17. A smooth curve is fit to the 1:5 and 1:17 mass ratio data to aid in identification of the various mass ratios. The average relative standard deviation (RSD) for all data was 9%, although error bars are not presented to avoid clutter [57].

PAGE 65

65 In a recent study, Tognoni et al. proposed a unique metric for determining plasma temperature that is independent of the electron density [62]. Assuming local thermodynamic equilibrium, a relationship between the ion-to-neutral ratios of two separate elements (here Mg and Cd) was derived as [ ] [ ] r -+--+ b t n n f b t n n f = Tk EEEEEE gA gA gA gA R RB Cd I k ion II i Mg I k ion II i Cd II i II ij I k I kl Mg I k I kl II i II ij III Cd III Mg)( )( exp/ / (3-1) where the following properties are defined: Aij and Akl are transition probabilities for the i-j (ion) and k-l (neutral) transitions, gi and gk are the statistical weights of the upper levels, Ei and Ek are the upper state energy levels of the ion and neutral species, respectively, Eion is the ionization potential, kB is the Boltzmann constant, and T is the plasma temperature. The subscripts refer to the elements Mg and Cd. In addition, correction terms are introduced to account for differences in instrument response at the two selected wavelengths for each element. These correction terms were set to unity, based on previous calibrations of the spectral response of our instrument. Finally, localized self-absorption is of concern if significant analyte mass is released from a single particle. Calculations were performed using the estimated atomic concentration based on complete particle vaporization, the estimated emitting volume about the aerosol based on previous measurements [18], and calculated absorption cross-sections. For all transitions, the maximum calculated turbidity was about 0.01, with the average value much less; hence all emitting volumes were considered optically thin. Turbidity values for both Magnesium and Cadmium transitions for different delays is shown in Table 3.1 and Table 3-2. In the present study, Equation 3-1 was implemented using the magnesium 280.27 and 285.21 nm ion and neutral lines, respectively, and the 226.5 and 228.8 nm ion and

PAGE 66

66 neutral cadmium lines, respectively. These lines are similar to the optimal lines suggested by Tognoni et al., with the exception of the 226.5 nm ion line, which was substituted in place of the 214.4 nm line utilized in their earlier study. However, the upper energy states of these two lines are nearly identical (44,136 vs. 46,619 cm-1), and comparative calculations revealed no significant differences in the algorithm. Energy Table 3-1. Turbidity values for Cd transitions for different delays Delay,s Central wavelength, nm Turbidity 2 226.09 0.01659636 5 226.09 0.00850906 15 226.09 0.00134058 25 226.09 0.00107825 30 226.09 0.00095761 40 226.09 0.00096336 2 228.39 7.459E-08 5 228.39 6.7432E-08 15 228.39 6.7385E-08 25 228.39 7.0521E-08 30 228.39 7.3182E-08 40 228.39 7.459E-08 Table 3-2. Turbidity values for Mg transitions for different delays Delay, s Central wavelength, nm Turbidity 2 280.0764 0.001735361 5 280.0764 0.000738986 15 280.0764 0.000142947 25 280.0764 0.000106921 30 280.0764 9.86645E-05 40 280.0764 9.69707E-05 2 285.0139 0.001807098 5 285.0139 0.000851833 15 285.0139 0.00016124 25 285.0139 0.0001165 30 285.0139 0.000108944 40 285.0139 0.000108144 values for Mg and Cd are tabulated in Table 3-3 .Overall, Tognoni et al. discuss the relative errors expected with this approach, and cites values in the range of 2%. Using

PAGE 67

67 Equation 3-1 for the lines detailed above, the calculated MgII:MgI/CdII:CdI ratio is presented in Figure 3.11 as a function of electron temperature. Table 3-3. Characteristics of Mg and Cd used in calculation of ion to neutral ratio Element Neutral line Ionic line Wavelength (nm) I k E (eV) Wavelength (nm) II i E (eV) Eion(e V) IIII iji II klk Ag Ag ft n n b Cd 228.8 5.417 226.5 5.47 8.993 0.4095 Mg 285.213 4.346 280.27 4.422 7.646 0.3502 Figure 3-11. Ratio of the magnesium ion-to-neutral (280.27 and 285.21 nm) intensity ratio to the cadmium ion-to-neutral (226.5-to-228.8 nm) intensity ratio as a function of plasma temperature based on the theoretical treatment of Equation 3.1 [57, 62] As presented in the original publication, the ratio of ion-to-neutral ratios is observed to decay monotonically with increasing plasma temperature, thereby becoming a useful metric for assessment of plasma temperature independent of the plasma electron

PAGE 68

68 density [62]. Using this temperature diagnostic in combination with the magnesium and cadmium emission data, the plasma temperature was calculated as a function of delay time for the three Mg:Cd mass ratios. The results are presented in Figure 3-12. Figure 3-12. Calculated plasma temperatures based on the magnesium-to-cadmium intensity ratios as a function of delay time with respect to plasma initiation based on the theoretical treatment of Equation 3.11 (see Figure 3.11). Representative error bars are calculated from the standard deviation of the measured experimental ratios. A smooth curve is fit to the 1:11 mass ratio data to aid in identification of the various mass ratios [57]. The data do present a degree of scatter, however, the overall trend is clearly one of an increasing temperature with increasing delay time. At the earliest times (2 to 15 m s), two of the three Mg:Cd mass ratios actually show a slight decrease in temperature,

PAGE 69

69 which is then followed by an increase at latter delay times. Because the overall bulk plasma temperature is decreasing with time from the onset of plasma initiation, the observed increase in local plasma temperature around the Mg-Cd particles is not perhaps so intuitive. However, if one considers finite heat transfer rates and the necessary energy to vaporize and dissociate the aerosol particles, then it is not unreasonable to assume that the local plasma temperature is initially suppressed in the vicinity of the aerosol, and therefore the local temperature will increase as heat is transferred from the surrounding bulk plasma. The trends observed in the present study are consistent with recent measurements reported by Niemax et al. in an inductively-coupled plasma [63]. In that study, SiO2 particles were found to display similar temperature profiles of an initial decrease followed by increasing temperature with plasma residence, with larger micronsized particles showing evidence of incomplete vaporization. Additional calculations are offered below in support of these comments. To gain insight into the vaporization and ionization of the aerosol particles, an energy balance was performed. Using the Mg and Cd mass fractions contained within an average aerosol particle size of 75 nm (estimated from the TEM analysis), the total particle mass is about 0.5 fg. This mass was then used in combination with the ionization and heat of vaporization energies (weighted average based on the particle composition) to estimate the total energy required to dissociate and ionize (50% ionization fraction) the aerosol particle, which yielded a value of 8x10-12 J. A region of plasma volume surrounding the particle was then coupled to supply this energy to the particle, and the corresponding reduction in temperature was calculated using the mass within this volume and the specific heat. The plasma specific heat was assumed equal to

PAGE 70

70 10 kJ/kg K at an average temperature of 10,000 K based on reported calculations [64]. Using a plasma volume equal to a diameter of 500 nm (about particle 6 diameters), these calculations predict a temperature drop of 7700 K. Clearly such calculations contain a number of assumptions and approximations, but nonetheless, are consistent with the concept of a localized perturbation of the plasma temperature. Furthermore, the relative effects of heat and mass transfer may be assessed in terms of the Lewis number ( a /D) which is a dimensionless parameter that compares the thermal diffusivity ( a = k/ r c) to the mass diffusion coefficient. Once again using the published data for the thermal conductivity [64], and our previously measured value of the diffusion coefficient for calcium atoms (D = 0.04 m2/s) [18], the Lewis number was calculated to be 0.25. Dalyanader et al. in their recent work calculated Lewis number numerically for Mg and Cd aerosol as a function of time and position in plasma [65]. Lewis numbers reported are on the order of unity and range from 0.67-4.48 which is well in cognizance with estimated values in this study. While approximations are inherent in this calculation, such a near-unity value fully supports the concept of finite time-scales for both heat and mass transfer, which is consistent with our previously measured rates of diffusion [18], and our current results of localized plasma temperature perturbations on the order of tens of microseconds. Given the temporal dependence of the problem due to finite diffusion timescales, the role of volatility can be considered further. The coupling of particle-derived mass to the local plasma perturbations is diminished with time as discussed above. Therefore, more volatile elements may enter the plasma earlier, leading to greater localized matrix effects, while less volatile elements will shift the vaporization rate to longer timescales

PAGE 71

71 where the overall effect is somewhat mitigated. Such comments may be considered analogous to the role of volatility in LA-ICP-MS, where competition between transport toward the MS skimmer and radial diffusion may play important roles in matrix effects. This framework is consistent with the overall trends observed in the Figs. 3.5 and 3.7 data regarding the effect of volatility of the added mass. In summary, the current results provide additional insight into the role of plasmaparticle interactions pursuant to quantitative aerosol analysis with laser-induced breakdown spectroscopy. Based on recent studies by our group and others, the current picture that is emerging concerns a complex interaction between the plasma gases and the aerosol particle, during which the finite time-scales of particle dissociation, and heat and mass transfer are fundamental processes (see Fig 3.13). The finding that the analyte emission derived from aerosol particles is affected by the presence of concomitant elemental fractions, as observed with the sodium and magnesium containing aerosol particles, is of significant importance to the LIBS community. Such findings are direct evidence of a matrix effect for aerosol particles, due primarily to perturbations in the localized plasma properties. However, such perturbations are minimized at longer plasma delay times, hence quantitative analysis should be performed with careful attention given to the temporal plasma evolution. In the larger analytical community, the effect of localized conditions about the individual particles is also relevant to the ICPMS and notably the LA-ICP-MS communities, where such effects are of increasing scrutiny with regard to the issues of elemental fractionation and matrix effects.

PAGE 72

72 Figure 3-13. Cartoon depicting localized plasma-particle interactions

PAGE 73

73 CHAPTER 4 ELECTRON DENSITY MEASUREMENTS btnStudy of electron densities and temperatures in plasma have been area of study for many decades. Knowledge of electron densities and temperature helps in understanding various instabilities in plasmas, plasma equilibrium, and various mechanisms including diffusion, energy transfer and so on. Stark broadening, Langmuir probe, Saha Boltzmann method, interferometer method, spectroscopy, are a few of the methods to measure these quantities in plasmas. All of them have their advantages and disadvantages. Thomson scattering is one of these techniques used to characterize plasmas and have been studied for decades, but its use in the LIBS community has been limited to negligible. A basic understanding of Thomson scattering is provided here. Thomson Scattering Thomson scattering is a well established analytical technique and has been used throughout the analytical community for plasma analysis including for measuring electron density, electron temperature and degree of ionization. However, its use specifically in LIBS research has been rather limited to date. Thomson scattering occurs when electromagnetic radiation interacts with charged particles in the plasma, namely free electrons, leading to acceleration of the free electrons and thereby emission of radiation that is referred to as Thomson scattering. Thomson scattering spectrum can be used to measure Te and ne without any assumption of LTE. Main advantages of using Thomson scattering to measure the plasma properties are

PAGE 74

74 No assumptions of LTE. Measurement can be radially resolved so that Abel inversion is not needed, and thereby no error in centre of spatially resolved measurement. The plasma is not perturbed. Temporal measurements possible with this technique. Localized measurement possible. Thomson scattering occurs from interaction of a probe laser with plasma density fluctuations arising from electron motions in hot and dense plasmas, when an electromagnetic wave moves in a plasma. In general, ionized gas leads to both Rayleigh and Thomson scattering. Rayleigh scattering is due to interaction of electromagnetic wave with atoms and ions, while interaction with free electrons gives rise to Thomson scattering. For incident radiation to move through a plasma and interact with free electrons, the plasma needs to be transparent for the wave so that the dynamics of the plasma can be probed using Thomson scattering. The condition for the plasma to be transparent to an incident probe wavelength is achieved when the probe frequency is significantly greater than the plasma frequency, as given by 1 2 2 4e p ene mp w ft = n b (4-1) where e is the charge of electron, ne is the electron density, me is the mass of electron. For example, for a plasma with electron density of 1x 1018 cm-3, the plasma frequency will be given by 5.6 x 1013 Hz. For this, the frequency of the probe laser needs to be order of 1014 Hz or more. Probe lasers of wavelength 1064 nm and 532 nm give frequencies of 2.8 x 1014 Hz and 5.6 x 1014 Hz, respectively. These wavelengths are sufficient for probing plasma of this high density, however at very early times in the

PAGE 75

75 plasma, high absorption is expected for this probe wavelength. With time, electron density decreases and so does the plasma frequency, and the plasma becomes more and more transparent to the probe laser and ultimately becomes totally transparent as shown previously by Hohreiter and Hahn et al[19]. Thomson Scattering Theory Incident wave with frequency b0 and wave vector k0 interacts with plasma density fluctuations with wave vector k ,and emits scattered radiation with wave vector ks ,which is detected at an angle t, as described by the k = sk 0k (4-2) Figure 4-1 Thomson scattering geometry [41] For non-relativistic case, it is a good assumption that incident and scattered wave will have same wavelength, so ( ) ( ) 0 04 2 2 2 kkkSinSin p q q l=@ = (4-3)

PAGE 76

76 Important parameters in Thomson scattering: There are two important parameters in Thomson scattering which define the characteristics of Thomson scattering and determine which characteristics of the plasma will be measured. They are briefly defined here: Debye length: The Debye length is the scale of spatial fluctuations in the fully ionized plasma, the minimum distance over which electrons shield the electric field. It is denoted by D l and is given as 1 2 24Be D ekT nel p ft = n b (4-4) where B k is the Boltzmann constant, Te is the electron temperature, ne is the electron density, and e is the charge of an electron. The parameter k as discussed previously, is the wave number of plasma fluctuations and depends on scattering geometry. 1/k basically defines the distance over which plasma fluctuations are probed. Thomson scattering parameter Dkl a1 = is defined as the distance of the plasma fluctuation probed divided by the plasma fluctuation distance. For k-1 << D l << 1, which means that the probed distances are smaller than the plasma fluctuation distance. In this limiting case, electrons behave independently and individual intertial electron motions are probed giving rise to a Gaussian profile which results from scattering from individual electrons. The profile is Doppler shifted from electron motions. is given by[41,66,67] 2 1 0) 2 (9.44 b t n n f =e eT n Sinxq p l a. (4-5)

PAGE 77

77 Figure 4-2. Regimes in Thomson scattering [67] 0 0.05 0.1 0.15 0.2 -60-40-200204060Intensity (a.u.)Wavelength (a.u.) a aa a >2 a aa a <<1 0.5< a aa a <2 Figure 4-3. Thomson scattering signal for different values of

PAGE 78

78 The parameter divides Thomson scattering into three regimesincoherent scattering, transition regime, and coherent scattering as shown in Figure 4-2 and 4-3. When << 1, scattering by an individual electron is measured and is incoherent in nature. As mentioned above, the profile is Gaussian in nature and the width of the spectrum may be used to provide the temperature of electron. All the scattering energy is concentrated in the electron Gaussian peak in this domain. Values of 0.5< <2, define transition regime. As shown if Fig 4-3 in this regime, the Gaussian profile has a flat top and electron shoulders appear. These electron shoulder profiles can provide both the electron temperature and density, which makes this regime very useful for plasma diagnostics. Location of the shoulder is determined by the electron density. In this regime, scattering energy is distributed between the electron peak and the weaker central ion acoustic peak. For >2, fluctuations probed are greater than the Debye length, and hence collective plasma fluctuations are probed in a process called coherent scattering. This regime has a low frequency ion peak and two symmetrically placed electron satellite peaks which are given by the Bohm Gross relation. The width of the ion peak is approximately given by the speed of sound in the plasma, and generally electron peaks are weaker in this regime. At very large values of alpha, electron peaks are weaker, farther separated, and extremely narrow having Lorentzian profiles. Ion peaks are much brighter, but because of the assumption of LTE, this peak cannot have resonance. However, for higher electron-to-ion temperature ratio (i.e.>>Ti),a high intensity ion peak is possible. The resulting scattered radiation, even though called ion peaks, are not exactly from ions. Actually, the scattered radiation is from the electron cloud which

PAGE 79

79 shields the ion. Figure 4-4 shows coherent and incoherent Thomson scattering resulting from the same plasma but with different values of resulting from different scattering detection angle Figure 4-4. Coherent and Incoherent Thomson scattering [67] For LTE, resonance occurs for a value of alpha >3.45, while for very large this value occurs at the plasma frequency. When Te >Ti, resonance occurs at the acoustic plasma frequency. For large and Ti, the main contributor is electron scattering, and the ion spectrum is neglected. The scattering cross-section is given by ),(),(wswskSkT=, (4-6) where T is Thomson scattering cross-section of free electrons and is given by 2 3 8e Tr p s= = 6.6 x 10-25 cm2. (4-7)

PAGE 80

80 S(k,) is called the dynamic form factor, which is a function of geometry and incident wavelength, and gives information regarding frequency shifts due to electron motions and other effects on electrons from ion motions, collisions, etc. Salpeter, in his derivation, beautifully separated the term into electron and ion terms, which better help in understanding the and interpreting the Thomson scattering and frequency shifts, and in turn measure the plasma properties [66]. Various assumptions were made by Salpeter while arriving at the final expression for the dynamic factor, which are listed below: Local Thermal Equilibrium Fully ionized gas No collisions 13>>=LDenl which implies a sphere of Debye radius contains many electrons and coulomb interaction between nearby electrons can be neglected as compared to thermal energy Certain limited deviations from LTE allowed which includes different electron and ion temperature, though difference cannot be very high. This leads to final expression of the dynamic form factor S(k,) which has two terms, one arising from ion contributions and the other arising from the electrons contributions[66,67]. (4-8) where xe = b/kve, xi = b/kve, and vi and ve are ion and electron velocities, respectively. The subscript i and e refer to ion and electron terms, respectively

PAGE 81

81 In laser induced plasmas, mostly conditions pertaining to > 1 are observed, hence those values are analyzed further. For large alpha, the electron spectrum consists of two resonance peaks located at 22 2 0 3(/) pBee kTmk ww=+ (4-9) There is a second peak given at xe = -/1.414. For ion term, for small ion term contribution is negligible. As increase, their contribution increases, but with the limit of electron-to-ion temperature ratio, no resonance occurs. Maxima occurs at xi = 1.5. When Te>> Ti then resonance occurs at ion-acoustic wave for large values of given by bac ( ) 2/1/ie acmZkTk =w (4-10) For this frequency, 1 2 1 2 1, 2 1. 2e e i e i iZm x m ZT x T ft = n b ft = n b r (4-11) Thomson scattering is perhaps the most useful for studying highly energetic plasmas despite the relatively low scattering cross-sections (~10-25 cm2). Therefore, the application of Thomson scattering for studying low-energy plasmas is challenging because of the presence of the high density of neutral species and reduced free electron density. Despite these challenges, the Muraoka and Hiftje groups have been successful in using Thomson scattering to study low temperature plasmas, including inductivelycoupled plasmas [41,68,69]. When >>1, the measured fluctuations increase and may exceed the Debye length and collective group motions are probed. These fluctuations are less damped as < >

PAGE 82

82 compared to the earlier case, and resonance occurs at the natural frequency of the plasma, referred to as the Bohm-Gross frequency for electron plasma waves as discussed by Villeneuve et al. [70]. The electron peaks are completely detached from central narrow ionic peaks (also called ion acoustic peaks), and the frequency of scattered light is both red and blue shifted from the incident probe frequency by an amount equal to Bohm-Gross frequency BG, as given by [70,71], sprBG www = (4-12) These resonances lead to non-elastic peaks in the Thomson scattering spectrum with frequency shifts given by BG w which are referred to as electron satellites. Here s w is the scattered frequency and pr w is the probe or incident frequency. In this phase, most of the scattered energy is from the ion feature. The resulting shape, intensity and separation of the Thomson profile and plasma frequency measurement can be used to measure electron density and electron temperature. As the value of increase, the intensity of ion peaks starts decreasing rapidly and measurement of ne and Te becomes difficult. The Figure 4.5 shows Thomson scattering for incoherent and coherent regime which is obtained by changing values of alpha by changing the scattering angle. For thermal non-relativistic plasmas like inductively coupled and laser-induced plasmas with high electron density, the low electron temperature incoherent regime is hard to achieve even by opting for various detection angles; hence mostly collective regime scattering is observed. Various detection angle schemes are applied, t< 900 (forward scattering), t= 900, t> 900 (backward scattering). Forward scattering gives larger alpha values with detached satellite and ion peaks, but suffers from stray light problems. Backward scattering detection angle can give very low alpha values and

PAGE 83

83 Gaussian profile thus obtained can be easily used for plasma analysis, but with very large detection angles again stray light is an issue. Hiftje and group report that detection angle beyond 135 degrees does not help much in Thomson scattering detection. Orthogonal detection angle (900) is very suited for experiments because of its easy implementation [41]. This detection scheme provides intermediate values of alpha with broader spectrum acquisition and can be easily used to measure plasma properties. While Thomson scattering can be used for measurement of plasma properties locally, it can also compete directly with the Rayleigh scattering signal, making quantitative analysis difficult. The Thomson scattered signal is very small compared to incident power and is generally in order of 10-9 10-11 times smaller than incident power. The small signal combined with stray light from surroundings, from the probe laser, and Bremsstrahllung radiation, can make the Signal-to-Noise ratio very small, hence these factors should be taken care of during the experiment design. The Thomson scattering cross-section is given by ) 1( 2 22 2qy sSinCos r d de-= W (4-13) where er is electron radius, y is angle between polarized plane of light and scattered wave, q is angle between incident laser probe and scattered wave. For unpolarized light this relation becomes )1( 2 2 2q sCos r d de+= W ) (4-14)

PAGE 84

84 Integrating over the entire solid angle of theta gives the total Thomson scattering crosssection as 2 3 8er p s= which is valid for both polarized and unploarized incident wave. For electron, this number comes to be 6.6 x 10-25 cm2. Compared to Rayleigh scattering for N2 (6.2 x 10-28 cm2/sr), the Thomson scattering cross-section is higher, but high density of neutral molecules as compared to electrons in a given plasma counters the high scattering cross-section of Thomson scattering. Stray light and background emissions further lowers S/N of Thomson scattering. Thomson scattering by ions is generally neglected owing to the heavy mass of ions compared to electrons. Experimental Setup As discussed above, it is shown that the measurement of electron density and temperature is very important in characterizing plasmas which can in turn help in better understanding various underlying processes and mechanisms of LIBS and related techniques like LA-ICP-MS. During very early times in the plasma, fractionation effects have been observed, as discussed in previous chapter, and transient electron densities at early times might play an important role in understanding this. Localized perturbations during early times in the plasma are related to the localized temperature and electron density which may differ from the bulk plasma conditions. In order to understand electron density and temperature at very early times in plasma, Thomson scattering and Stark broadening experiments were designed. The experimental setup is shown in Fig 45 [72]. Two different pulsed Nd: YAG lasers, one operating at 1064 nm (4 ns FWHM) and the second operating at 532 nm (5 ns FWHM), were used for these experiments. The

PAGE 85

85 1064-nm wavelength laser was used for creating the laser-induced plasma, while the 532nm frequency-doubled laser was used as a probe laser. Henceforth, these will be Figure 4-5. Top view schematic of the experimental setup for plasma scattering imaging and transmission experiments. referred to as the LIBS laser and probe laser, respectively. A single 100-mm focal length lens was used to focus the LIBS laser to the center of a six way cross that functioned as the LIBS chamber, thereby creating the laser-induced plasma in pure air under atmospheric conditions as shown in schematic in schematic above. Pure HEPA filtered air was used to purge the LIBS sample chamber at a continuous rate of 15 lpm for all experiments, thereby eliminating the presence of any aerosol particles and assuring a pure gas-phase breakdown for all experiments. The probe laser was expanded and collimated using a Galilean telescope setup, and then focused in the sample chamber orthogonal to the LIBS laser beam path such that the probe laser passed through the center of the LIBS plasma as shown in Fig4-4. Plasma emission and Thomson scattering

PAGE 86

86 were focused onto an ICCD camera using a UV grade achromatic lens, with the imaging axis parallel to the LIBS laser beam path (see Figure 4.4). A UV-grade substrate, 1064nm 45o dichroic mirror (not shown) was used to reject residual LIBS laser energy from the imaging ICCD. In addition, two 532-nm laser line filters (line width 8.5 nm) and additional polarizers were mounted in front of camera. The line width of 532-nm line filter was measured and characterized as shown in the Fig 4-6. 0.0 1005.0 1021.0 1031.5 1032.0 1032.5 103525 530 535 540Intensity (a.u.)Wavelength (nm) Figure 4-6. Line width of 532 nm line filter The polarizer was oriented to pass vertically polarized light with respect to the horizontal scattering plane, which corresponds to the polarization state of the incident probe laser. The flashlamp sync of the LIBS laser was used to trigger two delay generators, which in turn triggered both the flash lamp and Q-switch of the probe laser and also triggered the ICCD camera. Timing was adjusted in such a way that the probe

PAGE 87

87 laser and ICCD were synchronized (20-ns gate centered on the probe laser pulse), and could then be moved in concert relative to the LIBS laser pulse. For the experiments, the probe laser was moved relative to LIBS laser from 0 ns (i.e. coincident) to 500 ns following plasma formation. Hence for all experiments reported, a delay of zero corresponds to temporal alignment of the probe laser and LIBS laser peaks, and a 500-ns delay puts the probe laser 500 ns after the LIBS laser pulse (as measured peak to peak). Waveforms of both the lasers are shown here with delay of 10 ns peak to peak. The figure 4-7 also shows the detector 20 ns CCD gate which is centered on probe laser pulse. 0 20 40 60 80 100 120 140 160 -50.0-25.00.025.050.075.0100.0Normalized Intensity (a.u.)Time (ns) Detector gate 20 ns Delay LIBS Laser Probe Laser Figure 4-7. Waveforms of the LIBS laser pulse and the probe laser for a peak-to-peak temporal delay of 50 ns. The 20-ns intensifier gate width is synchronized to the probe laser, as shown.

PAGE 88

88 A fast response (200-ps rise time) detector and digital oscilloscope (2.5Gsample/s) were used to continuously measure and monitor the temporal delay between the two laser pulses for all experiments. For the current experiments, two LIBS laser pulse energies were investigated, namely 312 mJ/pulse and 500 mJ/pulse, referred to as the low and high energy pulse energies, respectively. The energy of the probe laser was fixed at 12 mJ/pulse for all experiments, which was insufficient to create a breakdown on its own. For the transmission experiments, the energy of the probe laser was measured after passing through the sample chamber in the absence of the LIBS laser to provide a reference, and then the probe laser energy was measured again after it passed through the plasma for various delay times. A laser power meter (Ophir Nova II) was used to measure the energy of the probe laser. All transmission measurements were repeated a minimum of three times, with the average transmission calculated as the direct ratio of transmitted energy through the plasma to the transmitted probe energy with no plasma. The transmission measurements were recorded for both the high and low LIBS laser energies over delay times from 0 to 500 ns with respect to the LIBS laser pulse. Thomson Scattering Imaging Plasma only images and images of plasma with the probe laser were recorded using an intensified CCD camera for different delay times ranging from 0 to 500 ns. Two laser energies for the LIBS laser were used312 mJ/pulse and 500 mJ/pulse. Typical scattering image grid size (10x10 mm) is shown in Fig 4-8. In figure 4-9, raw images of the plasma only (above) and of the plasma with scattering probe (below) are shown at 6 different delay times with respect to plasma creating laser. The arrow

PAGE 89

89 indicates the direction of the 532-nm probe laser. Scale bar shown in Figure 4.8 represents 1 mm. Figure 4-8. Typical Thomson scattering imaging. Grid size is also shown here which yields resolution of camera to be ~7.8m/pixel. Grid squares equal 1 mm. Figure 4-9. Raw images of the plasma only (above) and of the plasma with scattering probe (below) at different delay times with respect to plasma creating laser.

PAGE 90

90 Similar characteristics were observed for the high energy laser pulse with only minor differences as discussed below. As observed in all images (see upper images of each pair), there is a core plasma that grows larger with time. Superimposed on the plasma emission is a scattering effect from the probe laser which begins to appear with delay times of about 5 ns. The probe laser scattering is initiated and confined to the side of the laser-induced plasma facing toward the incident probe laser. The plasma scattering effect becomes stronger with increasing delay time to a maximum probe scattering at a delay value of about 15-20 ns delay. Following this maximum, the Figure 4-10. Processed plasma scattering images, in which the background plasma images have been subtracted from the plasma plus probe laser images. Set (a) shows scattering images from low energy experiments (313 mJ) at delay times of 10 ns (left) and 15 ns (right). Set (b) shows scattering images from high energy experiments (500 mJ) at delay times of 20 ns (left) and 25 ns (right). The rectangular box (25x25 pixel region) corresponds to the region used for calculating the scattering intensity data. Scale bar = 1 mm. (a) (b) 10 ns 15 ns 20 ns 25 ns

PAGE 91

91 scattering signal is observed to decrease with time, essentially vanishing completely by delay times of 90-100 ns. Figure 4-10 shows the scattering images following subtraction of the plasma-only emission (i.e. no probe laser) to provide a clearer picture of the scattering process. Together, Figures 4-9 and 4-10 demonstrate a very intense, transient scattering behavior with regard to the incident probe laser. To quantify the probe laser scattering, the scattering intensity was calculated from the images as a sum of all pixel values using the absolute scattering image (i.e. background plasma subtracted) for a 25 x 25 pixel box centered on the brightest portion of the scattering image. Representative boxes are shown in Figure 4.10 Figure 4.11 shows the resulting temporal evolution of the scattering effect for both the high and low energy LIBS laser pulse. The data reveal a clear maximum in scattering intensity in the delay range of 15-25 ns, with the more energetic plasma displaying the peak at slightly longer delay times. For both cases, the probe laser scattering signal is short-lived, namely, confined to delay times of less than 100 ns following plasma inception. 0.0 0.2 0.4 0.6 0.8 1.0 020406080100 Low Energy High EnergyNormalized Probe Laser Scattering Intensity Delay (ns) Figure 4-11. Temporal evolution of Thomson scattering as a function of delay time.

PAGE 92

92 For Thomson scattering spectral measurements, the ICCD camera was replaced by an optical fiber and the scattered light was fiber coupled (0.22 NA) to a spectrometer (Acton Spectra pro 300), where spectra were recorded using an ICCD detector with a 20-ns gate synchronized to the probe laser pulse. As discussed above, it is desirable to understand the evolution of the laser-induced plasma at early times in an effort to gain additional insight into the subsequent analyte response. The intense probe beam scattering discussed above is suggestive of an initial transient in free electron density resulting in significant Thomson scattering. To confirm the presence of Thomson scattering, additional spectral measurements were recorded using the fiber-coupled spectrometer as described above. The basic experimental setup remained the same as in Fig 4-5, except the ICCD camera which was replaced by an optical fiber, and the signal 0.0 1005.0 1061.0 1071.5 1072.0 1072.5 1073.0 1073.5 1074.0 107525 530 535 540Scattering Intensity (a.u.)Wavelength (nm) Figure 4-12. Spectral scattering profile of the probe laser for the high energy LIBS laser recorded at a delay of 25 ns

PAGE 93

93 was subsequently recorded using an ICCD array synchronized to the probe laser in the same manner as the imaging camera. Figure 4-12 shows the recorded spectrum of the probe laser scattering for a delay of 25 ns for the high energy LIBS laser case. The spectrum was recorded in a manner similar to the plasma images, namely, an average of 500 spectra was recorded for the plasma only (no probe laser) and then a second 500shot average was recorded with both the LIBS and probe lasers. The final probe laser scattering spectrum was the difference of the two. This approach eliminated the plasma emission that overlapped spectrally with the probe laser. Figure 4.-12 reveals a spectrum dominated by elastic scattering (mostly from stray laser light), making it difficult to observe the electron satellites or ion peaks characteristics of indicative of Thomson scattering. However, a closer examination of the Figure 4-12 data is presented in Figure 413, in which the two equidistant peaks are readily apparent which is suspected to be electron satellite peak, but it needs further investigation to confirm or reject those peaks as electron satellite peaks. The two peaks occur 2.7 nm from probe laser wavelength of 532 nm. With stray light dominating the spectrum, there is a possibility that those two peaks may be result of an artifact. However, while the current imaging measurements confirm the presence of significant Thomson scattering, the spectral data are characterized by rather poor signal-to-noise ratios, and do not provide any significant information regarding electron or ion peaks. For quantitative analysis, it is desirable to use a more sophisticated optical set-up such that increased spatial resolution and straylight rejection are realized, thereby producing a much greater signal-to-noise ratio.

PAGE 94

94 4 1048 1041 1052 1052 105528.00530.00532.00534.00536.00Scattering Intensity (a.u.)Wavelength (nm) 2.7 nm 2.7 nm Figure 4-13. Close-up of the Thomson scattering spectrum presented in Figure 4-12 Figure 4-14. Further zoomed Thomson scattering spectrum

PAGE 95

95 N Line Stark Broadening In order to explain and understand the results of the Thomson scattering imaging experiments, electron density measurements are helpful. To determine the electron density at very early times in the plasma, experiments were performed to measure the full-width-half-maximum of an appropriate N line. The experimental setup for these measurements has been discussed in previous studies [19,53]. In this study, the electron density was calculated by measuring the full-width-half-maximum of N(II) line at 500.1 nm (166,582.45 cm-1 -186,570.98 cm-1) at different delay times. This line, while not extensively studied and used in research community, is prominent at very early times in the plasma ( as early as 5 ns) and is accompanied with significant broadening, making it a suitable candidate for electron density estimation via Stark broadening. The measured line profile was fitted with a Lorentzian curve to determine the full-width-halfmaximum widths. Figure 4.15 shows original profiles as well as fitted profiles at two different time delays, which clearly show change in line widths with time. -2 1030 1002 1034 1036 1038 1031 1041 1041 104494.00496.00498.00500.00502.00504.00506.00508.00Intensity (a.u.)Wavelength (nm) 170 ns 30 ns Figure 4-15 N(II) line broadening and fitted Lorentzian profile

PAGE 96

96 This N(II) line is a result of three peaks at wavelengths of 500.1 500.27 and 500.515 nm. Lorentzian profiles were fitted for each of these wavelengths with the same width, such that the sum of these three lines matches the measured profile. Equal widths were assigned to each since energy levels for each transition were very similar, as listed in Table 4.1. Figure 4.16 shows three fitted profiles for triplet at 200 ns delay time. Table 4-1. Spectral properties of N(II) lines [73] Wavelength (nm) Ei -Ek (cm 1 ) gi gk Aki (s 1 ) 500.15 166,582.45 186,570.98 5-7 1.05e+08 500.27 148,908.59 168,892.21 1-3 8.45e+06 500.51 166,678.64 186,652.49 7-9 1.16e+08 Figure 4-16. Lorentzian fitted profile of three N lines As necessary for Stark broadening measurements, the line width is corrected for instrumental profile width which was calculated to be 0.13 nm using a low pressure Hg lamp as shown in Figure 4.17. For these measurements, Stark broadening parameters

PAGE 97

97 Figure 4-17. Instrument profile correction using measured line width of Hg lamp of Griem at 25,000 K were used, which is a reasonable assumption of plasma temperature at early times in plasma [43]. Electron densities can be calculated from these parameters by following relation: *enC l =D (4-15) where C has been tabulated by Griem, as summarized in Table 4.2, and l D is the full width of the line in nm. Table 4-2. Stark broadening parameter for Nitrogen 500.1 nm Temperature (K) C 5000 2.82486E+18 10000 3.50877E+18 20000 4.03226E+18 40000 4.23729E+18

PAGE 98

98 Figure 4-18. Griems parameter plotted against Temperature. Third degree polynomial is fitted to obtain values of C at intermediate temperatures. These values were then plotted against temperature and a third degree polynomial is fit to interpolate the values of these parameters at a temperatures of 25,000 K (Figure 4-18). Interpolated values are tabulated in Table 4-3, which were used for calculations of number density of electrons at different temperatures, as listed in Table 4-4. Table 4-3. Griems parameter calculated for different temperatures by curve fitting Temperature (K) C as calculated from fitted polynomial 5000 3.27E+18 10000 4.12E+18 17500 3.95E+18 20000 4.97E+18 25000 4.09E+18 32500 4.17E+18 40000 5.73E+18

PAGE 99

99 Table 4-4. Electron density as a function of delay time Electron Density Calculation Corrected Delay Corrected Line Width Ne @ 25000 K Ne @ 17500 K Ne @ 32500 K 5 5.17 2.11E+19 2.04E+19 2.16E+19 20 5.12 2.09E+19 2.02E+19 2.14E+19 30 4.67 1.91E+19 1.84E+19 1.95E+19 40 4.37 1.79E+19 1.73E+19 1.82E+19 50 4.07 1.66E+19 1.61E+19 1.70E+19 60 3.77 1.54E+19 1.49E+19 1.57E+19 70 3.42 1.40E+19 1.35E+19 1.43E+19 90 2.94 1.20E+19 1.16E+19 1.23E+19 110 2.37 9.69E+18 9.36E+18 9.88E+18 130 2.07 8.47E+18 8.18E+18 8.63E+18 150 1.85 7.57E+18 7.31E+18 7.71E+18 170 1.72 7.03E+18 6.79E+18 7.17E+18 195 1.55 6.34E+18 6.12E+18 6.46E+18 220 1.37 5.60E+18 5.41E+18 5.71E+18 320 1.02 4.17E+18 4.03E+18 4.25E+18 As shown in Figure 4-19, highest electron density estimated by Stark broadening is 2.1 x 1019 cm-3 at 5 ns delay time from plasma inception, which decays to ~1 x 1019 cm-3 by 100 ns and to 5.6 x 1018 cm-3 by 220 ns. Parriger et. al. report electron densities of ~ 1x 1019 cm3 and ~1.2 x 1018 cm-3 at 5 ns and 100 ns respectively [20 ]. Compared to Parrigers data, the electron densities estimated here are about two times higher at 5 ns and about an order of magnitude higher at 100 ns delay time. Comparing to our previous study these values at 100 ns are about 5 times higher, which can be attributed to uncertainty in values of Stark broadening parameters of the N line used for electron density calculation, and accuracy of line width measurements .Nevertheless, these calculations give a good estimate of the transient electron density in the plasma from as early as 5 ns.

PAGE 100

100 5.0 10181.0 10191.5 10192.0 10192.5 10190 50 100 150 200 250Electron density (cm-3)Dealy (ns) Figure 4-19. Transient electron density at 25,000 K calculated via Stark broadening Interpretation of Thomson Results and Electron Density Measurements: The Thomson scattering parameter, was calculated by using the electron density as obtained from the Stark broadening measurements. Electron temperature and ion temperatures were assumed to be 30, 000 K (LTE assumed). The parameter is plotted against electron density in Figure 4-20. For the electron density range, observed in this experiment, it is seen that the value of >1. For the electron densities of order of ~1019 cm-3, which corresponds to delay time of 5100 ns, >>1. Based on Salpeters treatment of Thomson scattering, two electrons satellite peaks are expected [66]. Evans et al. have showed that the for >2, satellite peak separation from the laser wavelength

PAGE 101

101 15 20 25 30 1.0 10191.5 10192.0 10192.5 1019alpha ( a a a a )Electron Density (cm-3) 5 ns 90 ns plasma decay time Figure 4-20. Thomson scattering parameter for different electron densities is given by [66] 2 14 32 2.1101.1310sin() 2 s e e xnxT q l-D=+, (4-15) where Te is in ev and l D is in Angstrom units. For ne = 2 x 1019 cm-3 Te = 30,000 K= 2.5 ev, t = 450, l D = 64.9 nm At such high values of resonance electron peaks are predicted, separated by distance of ~40-60 nm, as calculated above (assuming LTE), while for ion peaks no such resonance is predicted. In the current study, for imaging experiments, 532 nm line filters were used with FWHM of 8.5 nm. Thus, electron peaks, even if present, will not be detected by the camera. So, it is confirmed now, that for the range of in the current study, satellite

PAGE 102

102 peaks cannot be detected by the imaging ICCD. The two equidistant peaks, discussed earlier, can be dismissed as an artifact or noise and need not be considered further. This raises additional questions regarding the observed intense scattering. Ion peaks cannot resonate under LTE assumptions, but if LTE is violated and Te>> Ti, resonance peaks from ions can occur, which are also referred to as acoustic peaks. This discussion provides a plausible explanation that the Thomson scattering may be due to the resonance from the ion peaks, which thereby indicates presence of non-equilibrium conditions very early in the plasma. With increasing delay time, plasma comes to equilibrium resulting in a significant decrease in the intensity of the ion scattering, and possible increase in the intensity of electron peaks (which cannot be detected in this study). These results show that, even when the electron density is not decreasing rapidly, the presence of non-LTE conditions may give rise to intense Thomson scattering from ion peaks which decays as the LTE condition is reached. Ion peaks may still be present even after LTE is reached, but the intensity will be extremely diminished (for >>1). Based on this plausible explanation, it can be inferred that the plasma reaches LTE (Te= Ti) by a delay of ~20-40 ns after the inception of the plasma. Thomson scattering signal after this delay can be attributed from non-resonating ion peaks. Transmission Experiments To gain additional insight into the early plasma dynamics, transmission measurements were also performed to understand how much of the probe energy was being attenuated by the plasma, presumably via Thomson scattering and free-free absorption. The discussion in this section follows from the published work of Diwakar and Hahn. From the electron density measurements, plasma frequency is estimated to range from 4.1x10132.1x 1013 Hz during 5ns to 220 ns delay time. These frequencies

PAGE 103

103 are significantly close to probe laser frequency (5.6 x 1014 Hz) thereby suggesting significant absorption and attenuation during these delay times. Figure 4-20 presents the measured probe laser transmission for the high energy LIBS laser. The transmission plot clearly shows that most of the probe laser energy is absorbed initially, with the laserinduced plasma essentially opaque to the probe laser for delay times of 10-20 ns following plasma initiation. This is consistent with previous measurements in a similar but less energetic plasma [19]. In Figure 4-21, the minimum transmission value of about 6 % occurs at a delay of 10 ns, which correlates well with the maximum in the Thomson scattering data. Such a correlation suggests a strong coupling of the probe laser to the free electrons during the initial plasma transient, although it is not possible to partition the contributions of free-free absorption and Thomson scattering to the overall probe beam attenuation during the periods of maximum electron density. However, at longer delay times, additional comments are offered. As the free electron density decreases with delay time, the Thomson scattering becomes markedly weaker, as observed in the Figure 4-11. By 100 ns delay, the Thomson scattering is essentially negligible as compared to the peak values, while the probe laser transmission is still less than 50%. This data suggests that the role of free-free absorption remains important, while attenuation via Thomson scattering is most likely restricted to temporal regions of the highest free electron densities which are possibly due to non-equilibrium conditions. At longer delay times, the plasma becomes dissipative and optically transparent, and is totally transparent by 500 ns, at which point the plasma frequency has become less than the probe laser frequency (~5.6x1014 Hz), as previously discussed [19].

PAGE 104

104 0.0 0.2 0.4 0.6 0.8 1.0 0100200300400500Transmission Delay (ns) Figure 4-21. Measured transmission of the probe laser beam through the plasma as a function of delay time for high energy LIBS laser. The delay of zero corresponds to plasma initiation and a peak-to-peak overlap of the LIBS and probe lasers. Error bars are one standard deviation. Investigation of Laser Breakdown in Gases Breakdown of gases by a laser pulse is an interesting phenomenon where photons interact with the matter converting a non-conducting, transparent gas to an optically thick, conducting plasma. Laser-induced breakdown in gases has been studied extensively since the year 1963 when Maker et al. reported the breakdown process in air and explained the process experimentally and theoretically [74]. Morgan et al.

PAGE 105

105 developed a detailed theory explaining various processes and mechanisms in the breakdown of gases [75]. Understanding the breakdown process including initiation and temporal progression of plasma formation is very important for full understanding of the plasma chemistry and thereby plasma-particle interactions. Breakdown of gases by a laser can be divided into four stages, namely initiation, formative growth, plasma growth and extinction. During the breakdown initiation process, multiphoton ionization provides the free electron for the breakdown to begin, which is followed by the growth in the number of electrons and ions by a cascade breakdown process. Both the mechanisms have been discussed in chapter 1. Following the initiation process, during the formative growth stage, the electron density increases and reaches a critical limit and breakdown occurs. The initiation and formative growth process lasts for only a few nanonseconds. During the formative growth stage, most of the laser pulse energy is absorbed in the process. The threshold for optical breakdown is described as laser power at which half of the laser pulse induces breakdown in the gas [75,76]. Chen et al. have discussed that not all of the incident laser energy is used for breakdown of gases, only a fraction of the incident energy[77]. The Fig 4-22 shows energy absorbed in the laser induced breakdown by two laser profiles with peak energies of 1 and 2.5 times the breakdown threshold, respectively. Shaded area shows the amount of energy absorbed in the process. No breakdown occurs until the laser pulse energy reaches the breakdown energy, and so the leading edge of the laser pulse will not be absorbed. The higher the energy of the laser pulse, as compared to the breakdown energy, the earlier the initiation process takes place and has less variation in the plasma initiation time [76]. The variation in plasma intuition time can be 15 ns, being less for high energy pulse energy.

PAGE 106

106 Figure 4-22. Energy deposition in laser induced breakdown by lasers having peak energies 1 and 2.5 times the breakdown threshold energy [77]. After the plasma breakdown process, the plasma growth stage takes place, which includes an increase in the plasma temperature, pressure, velocity and shock wave formation around the plasma. Saturation in plasma absorption occurs at a laser pulse energy three times the breakdown threshold value. In such cases, the trailing edge of the laser pulses does not increase the temperature of the plasma, but rather causes the plasma to expand. Chen et al. have reported a velocity of the plasma in the range of 80 Km/s for the first few ns for a laser pulse energy of 80 mJ [77].The plasma propagation velocity decreases to 20Km/s for rest of the remainder of the pulse. The plasma growth process is followed by the extinction phase which lasts for 100s of microseconds. Plasma extinction occurs due to radiative cooling, removal of electrons by diffusion, and recombination processes. The most important process in the breakdown of a gas by the laser pulse is the initiation process, which provides the first electron for the cascade process. Any factor which affects the initiation time and location will ultimately affect the breakdown process and in turn affect the plasma chemistry. The important parameters which can

PAGE 107

107 affect the initiation process are the laser pulse energy, focal length, spherical aberrations, and environmental factors including humidity and presence of aerosol particles. Chen et al. have shown that the laser pulse energy plays a very important role in breakdown process. A high energy laser pulse decreases the variations in the plasma initiation time and the location of plasma initiation is affected with the laser energy, as a high energy pulse leads to plasma initiation closer to the laser away from the focal plane. Spherical aberration in the lenses causes a multi point plasma initiation process. Chen et al. have shown that the environmental factors like humidity and temperature do not affect the breakdown process, while the presence of aerosols can certainly affect the breakdown. In order to study these mechanisms and understand the plasma inception process, additional experiments were conducted, with the goal of understanding the plasma chemistry at very early times in the plasma. The experimental set up for this is described in detail later in chapter 5.A pulsed Nd:YAG laser (1064 nm, 8ns FWHM, 5 Hz, 400 mJ/pulse) operating at its fundamental wavelength was used for creating the plasma, while the bottom of the six-way cross LIBS chamber was connected to the flow meters for supplying different gases in the chamber. Three different gases namely nitrogen, argon and helium were used at flow rate of 10 lpm in the LIBS chamber. The LIBS laser was focused onto the centre of six-way cross LIBS chamber using a 100 mm focallength condensing lens. Plasma emission was focused using UV grade achromatic lens onto the ICCD2 camera (Princeton Instruments). A series of images were recorded for different delay times. Fig 4.23 shows the temporal laser pulse profile in the current study.

PAGE 108

108 Figure 4-23. Temporal laser pulse profile used in current study The delay time zero corresponds to the time when the first breakdown process was observed. Corresponding to this zero delay time, images were also recorded for 10, 20, 30 and 40 s. Fig 4-24 shows the breakdown images for different gases for different times. As can be seen, at zero delay breakdown process starts as multiple spots which can be attributed to spherical aberration in the lens. The multiple plasma initiation process is more pronounced in nitrogen environment, where the number of plasma initiation points are more as compared to both helium and argon, and also well separated. Fig 4-25 shows the affect of spherical aberrations on the multiple plasma initiation process. The plot shows the isophyte lines which are the lines of constant intensity plotted around focus of a Gaussian beam. The first plot (a) is for an optical

PAGE 109

109 Figure 4-24. Breakdown images for Nitrogen, Argon, Helium at delay times of 0, 1, 10 and 20 ns after the plasma inception. Laser direction is from left to right. Figure 4-25. Lines of constant intensity, isophytes plotted for optical systems with (a) no aberration and (b) high degree of aberration [75] system with zero spherical aberration while (b) is the plot for the optical system with high degree of spherical aberration. It can be observed in the plot, that spherical aberration leads to loss in symmetry of the isophyte lines around the Gaussian focus.

PAGE 110

110 Also it leads to an increase in energy distribution along the axis of the laser beam, away from the focal plane towards the laser beam. Morgan et al., have reported that the larger the aberration, the wider is the energy distribution, which can be as wide as 1 mm along the laser beam axis providing multiple locations for the breakdown initiation process. As can be seen from the breakdown images, that for all three gases the breakdown process starts at multiple points and later in time, these multiple plasmas coalesce together to form a larger and more stable plasma. The degree of separation and location of these initial multiple plasma is different for different gases which needs to be investigated. The difference in the breakdown process for different gases can be attributed to difference in breakdown threshold energy levels for different gases. The shape and the location of plasma during plasma growth process depends on various energy transfer mechanisms including losses. Self-focusing for breakdown in nitrogen has been observed by Bunkin et al. which results in discrete breakdown structures, which can probably explain the observed large number of plasma initiation locations in the case of nitrogen [78]. Electronic polarizability of Nitrogen and air plays an important role in the self focusing mechanism, where during the first stage polarizability of the medium is enhanced, leading to a second stage where significant changes in refractive index occur, ultimately leading to a self focusing even at electron densities smaller than breakdown threshold. Self-focusing contributes mostly to later stages of plasma growth process by providing excited atoms and molecules. Self focusing mechanism has been argued and debated by many researchers whether it really affects the breakdown in gases or not. While some argue it does not affect the plasma initiation process too much, others have shown in studies involving nitrogen and helium that self focusing occurs and affects the

PAGE 111

111 initiation step of plasma formation [79,80,81] Differences in plasma shape for different gases later in time can be explained by understanding various losses and recombination mechanisms. Morgan et al. have discussed that for He, the recombination coefficient is very small as compared to argon, and hence no diffusion or recombination losses was observed (R< 2 x 10-11 cm3 s-1 at 1800 K). For argon, the recombination coefficient values have been reported to be 2x 10-8 cm3se-1 at 1800K. Further investigation needs to be done to clearly understand the complete breakdown process and plasma propagation. This preliminary study gives basic insight into the various mechanisms and factors affecting the breakdown initiation and plasma propagation, including spherical aberration, breakdown threshold, self-focusing, and recombination losses and so on and further study needs to be done to understand it. Conclusion The current study explores the initial evolution of the laser-induced plasma with a goal of better understanding the processes taking place which may have an impact on LIBS measurements. Previous studies have shown that analyte dissociation, heat transfer, and mass diffusion are finite processes rather than instantaneous events [18,65]. It follows that local temperatures and plasma conditions (i.e. in the vicinity of the analyte species) may be different than the bulk plasma properties, notably so at earlier stages of plasma evolution, which can affect the LIBS analyte signal in the form of matrix or fractionation effects. The significant observed Thomson scattering during the initial plasma evolution shows the presence of a highly transient electron density along with non equilibrium conditions, which may ultimately have implications on the resulting analyte emission. Stark broadening of N line gives a good estimate of electron densities at very early times in the plasma and the data is unison with Thomson

PAGE 112

112 scattering observations. Preliminary breakdown studies underline the importance of understanding the breakdown process at very early time in the plasma for different gases, which can ultimately have impact on plasma chemistry and thereby plasmaparticle interactions. Ultimately, it is believed that a better understanding of the fundamental plasma processes is a key to improve and develop LIBS as an accepted analytical method. Key Observations from this study are listed as follows: Electron density in the order of 1019 cm-3 was measured as early as 5 ns delay in the plasma. Transmission measurements revealed that the plasma was opaque for delay times of 10-20 ns which correlates with high electron density measurements and Thomson scattering observations. At longer delays, optically transparent plasma was observed. Plausible explanation for Thomson scattering is due to the occurrence of ion resonance peaks which is attributed to non-LTE conditions at early times in the plasma. Thomson scattering experiments suggest non-equilibrium conditions during the first ~10-20 ns after plasma inception. Difference in the breakdown mechanisms were observed for different gases namely nitrogen, argon and helium at very early times in the plasma which can be attributed to spherical aberrations, difference in breakdown threshold self-focusing and recombination losses.

PAGE 113

113 CHAPTER 5 SINGLE DROPLET INVESTIGATION IN LIBS Introduction Understanding the plasma-particle interactions and the related fundamental processes involved is important in advancing laser-induced breakdown spectroscopy as a reliable and accurate analytical technique. Critical understanding of the plasma-particle interactions include study of the plasma evolution, analyte atomization, particle dissociation and diffusion. Analyzing single droplets and particles in the analytical plasmas can provide great insight into the various fundamental processes, namely vaporization, desolvation, atomization, plasma-particle interaction and so on. In the experiments performed until now, the sample analyte contained a large number of the particles, and the resulting understanding from such experiments correspond to complex interactions of all the fundamental processes taking place. A nebulizer produces hundreds and thousands of droplets of varying sizes in the plasma volume per second and hence single droplet study is not possible in such an arrangement. Decoupling these fundamental processes can provide a forward leap in understanding the analytical plasmas with implications to LIBS, ICP-MS, and ICP-AES. Study of single droplets and particles temporally and spatially will provide such an opportunity of decoupling various fundamental processes, and to study their effects on analyte response individually. A key assumption for LIBS is that complete dissociation of the constituent species within the laser-induced plasma results in independence of the analyte atomic emission signal on the analyte source. This assumption has been challenged by recent findings, especially for short delay times, which are attributed to localized interactions [18,57]. Local analyte temperature may not be the same as the bulk plasma temperature, which can

PAGE 114

114 have implications when calibrating the analyte signal. The present study examines the plasma-particle interaction by tracking the temporal evolution of the analyte signal from single particle events by measuring the temperature using the Boltzmann plot method. Imaging studies are done to visualize the spatial location of particles inside the plasma volume and to estimate the diffusion rates of analyte species inside the plasma. In many cases, samples are introduced into the LIBS plasma as liquids (i.e. aqueous solutions). Many studies have shown that liquid analytes can be analyzed using LIBS efficiently, but it suffers from low sensitivity owing to inefficient plasma propagation in the liquid. The problem also arises when the sample available for analysis is limited (say < 1 ml). Several reasons can lead to limited availability of sample including biological experiments, forensics, clinical chemistry, toxicology, experiments involving radioactive analytes, speciation analysis etc. Collection of samples in the range of 10-50 nl is very common in above mentioned fields, and reliable techniques for the introduction of the sample analyte is required [82]. Increasing the volume of sample by diluting can be a part of the solution, but this can lead to many sensitivity issues. So the most adequate and appropriate sample introduction system for such studies should be discrete, involve low consumption rate, and provide high sensitivity. A sensitive, robust and reliable introduction system for analyte introduction in the nanoliter-picoliter range is discussed, and the detection limits using LIBS is determined here. Picoliter range analyte introduction in the plasma can have an advantageous effect, as the plasma cooling will be limited to localized region as compared to the analyte introduction using a nebulizer, where heavy mass loading can lead to overall plasma cooling and thereby interfere with the LIBS signal. A liquid droplet containing low volumes of analyte can

PAGE 115

115 address the above mentioned issues pertaining to small sample availability, and sensitivity issues in aqueous media. A single droplet provides a better means of analysis by LIBS as the breakdown mechanism in an isolated droplet is different than in bulk liquids. Studies have shown that in isolated droplets breakdown occurs outside the droplet in the forward direction, and then it is propagated back by a shock wave resulting in plume formation and material ejection from the droplet [83]. Considering these objectives, the following experiments were carried out in this work: Using a scientific-grade droplet generator, single droplets of analyte were generated and with sample volume in the range of sub nanoliters-picoliters, which were then introduced in the LIBS plasma. The second goal was to obtain temporal and spatial characteristics of plasma dynamics and analyte dissociation in conjunction with the analyte response. Plasma and particle imaging was used to estimate the diffusion coefficients of various analyte species, including calcium, hydrogen and gold. Multi-element analyte compositions were examined, including elements with both intense neutral and ionic emission lines (Mg/Ca). The goal is to calculate the local plasma temperature. Such measurements provide information regarding the stability of the plasma, non-equilibrium conditions, and the related analyte response. The overall goals are to study further the previously observed perturbations in localized plasma conditions in vicinity of the analyte, and how such effects are coupled to the LIBS analyte response. Experimental Setup Single Droplet Sample Introduction System A scientific grade microdrop dispenser system was used for generating single droplets of analyte in the size range of 40-60 m (MD-K-150, Microdrop Technologies). Hieftje et al were one of the first groups to use such droplet generator system single droplet studies inside the flames [84].Various research groups have used a similar kind of droplet generator and droplet introduction methods in LIBS/ICP plasma [84-87]. It is important to mention here that while in these studies scattering from droplets was used

PAGE 116

116 to actively trigger the laser pulse, while in the current study the droplet generator is synced with the laser using delay generators .The system consists of a dispenser head, a reservoir and an electronic controller. The reservoir can hold about 5 ml of liquid. The dispenser system uses a piezoelectric nozzle, commonly used in ink jet printers, to generate on demand single isolated droplets. The system uses a capillary tube, surrounded by a piezoelectric tube, which is filled with the sample liquid and the piezoelectric is activated with a voltage pulse which sends a pressure pulse across the liquid causing liquid motion at the nozzle. The velocity of the liquid reaches several meters per second in short time, and then it decelerates owing to the pressure loss and expansion through the nozzle. Combined with inertial forces and inertial surface tension forces, individual single droplets are formed at the nozzle propagating at a speed in the range of 1-4 m/s. The repetition rate can be chosen from 1-2000 Hz. The size of the droplet is dependent on the nozzle diameter, viscosity of liquid, and control parameters, including applied voltage and pulse width. Manufacturers of the system have reported reproducibility of the droplets to be very high if all the parameters are kept optimum and constant. They report variability in droplet volume to be less than 1%. Several other studies using similar droplet generation system have reported variability to be in range of less than 1-5% [88,89]. The droplet generation process can be triggered internally, externally, or manually. Strobe light arrangement is used for visualizing the droplet formation at the nozzle, where the delay of the stroboscope ranges from 0-900s. In the current study, a 30 m nozzle diameter is used for generating the single isolated droplets. The resulting droplets are in range from 40-60 m with exit velocity ranging between 1.5-2 m/s. All the experiments in the current study were performed

PAGE 117

117 with a droplet size of 50 m, which is equivalent to analyte volume of 65 pL. Fig 5.1 shows a single droplet generated by using the Microdrop droplet dispenser system in this study. Droplets were generated using the external trigger mode wherein the flashlamp of 1064 nm LIBS laser was used to trigger the droplet generator with a frequency of 5 Hz. Table 5-1. Droplet generator parameters Droplet generator model : MD-K-11150/176 Nozzle Diameter: 30 m Connecting tube: PTFE Driver Voltage: 7090 V Pulse length: 40-50 s Drop speed: 1.5-2 m/s Droplet diameter: 50 m Repetition rate: 5 Hz, External Trigger Figure 5-1. 50 m single droplet of analyte generated using a scientific grade droplet generator

PAGE 118

118 Optical and Data Collection Setup Two different Nd:YAG pulsed lasers, one operating at its fundamental wavelength (1064 nm, 5ns FWHM, 5 Hz, 315 mJ/pulse), and other operating at 532 nm (5 ns FWHM, 5 Hz, 15 mJ), were used. The 1064 nm laser was used for creating the laserinduced plasma, while the 532 nm laser was used for droplet imaging/visualization and droplet alignment in the LIBS plasma. The LIBS laser is focused onto the center of the six-way cross LIBS chamber using 100-mm focal-length condensing lens. The imaging laser is focused in the LIBS chamber such that it coincides with the hottest spot in the plasma spatially. Plasma emission is collected by employing backscattering geometry using a pierced mirror, and the subsequent signal is fiber coupled to a spectrometer (Acton Spectra pro 300), and finally to the ICCD camera (#1) (Princeton Instruments). Plasma emission was focused using UV grade achromatic lens onto the ICCD camera (#2) (Princeton Instruments or Andor). Single droplets of size 50-m were introduced into the chamber using a commercial scientific-grade droplet generator system (Microdrop Technologies). The flashlamp sync of the LIBS laser was used to trigger two delay generators, which in turn triggered, the flash lamp and Q-switch of the 532 nm laser, and also triggered ICCD camera1 and the droplet generator. The Q-switch of the LIBS laser was used to trigger the imaging camera, ICCD2. Timing was adjusted in such a way that the imaging laser and two ICCDs could then be moved in concert relative to the LIBS laser pulse. The droplet generator timing was adjusted in such a way that the droplet is temporally and spatially aligned with the LIBS plasma. Fig 5.2 shows the schematic of the experimental setup, while Fig 5.3 shows the front view of the zoomedin view of the droplet generator, and the LIBS chamber. The Figure 5.3 also shows the droplet generator setup and the image of the single particle in the plasma, as obtained

PAGE 119

119 from imaging using 532 nm laser. A nitrogen co-flow of 0.2 lpm was used to drive the droplet in the plasma. The size and velocity of the droplet was measured by calibrating Fi g ure 5-2. Schematic of experimental setup for single droplet experiments Figure 5-3. Front view of the schematic of experimental setup for single droplet experiments

PAGE 120

120 the nozzle diameter as observed on the monitor screen. To prevent capillary clogging, the analyte was filtered using 0.2 m pore membrane filter before transferring the analyte to the reservoir. Also regular cleaning of the capillary was done with ultrapurified deionized water and 3% HNO3. Calibration Curves and Limits of Detection for Ca and Au Ca and Au were used as analyte species for obtaining the calibration curves and detection limits using the single droplet analyte introduction approach in LIBS system. The analyte solution was prepared by diluting ICP-grade analytical standards of 10,000 g/ml of Ca and 1,000g/ml of Au (SPEX Certiprep). The standard solutions were diluted to the desired concentration using ultrapurified deionized water and stock solutions were prepared in range of 0.5-10 g/ml for Ca, and 100-500g/ml for Au. Single 50-m droplets were generated using a scientific grade droplet generator as described in the previous section, and injected into the plasma from a distance of 10 mm above the plasma. Representative Au spectra for a concentration of 1000g/ml are shown in Fig 5.4. The Figure shows three different Au spectra obtained from the same concentration of 1000 g/ml, which depicts the variation of LIBS spectra obtained from single droplets. For the current study, a series of 100 sets of such spectra were recorded and the peak-to-base ratio (P/B) was calculated for the Au atomic emission line at 267.59 nm (037,358.991cm-1). All the spectra with P/B value smaller than 10 were rejected, as those spectra contained essentially noise, and were results of droplets missing the plasma. All the remaining spectra after this screening process were ensemble averaged to give an average P/B value. This process was repeated for three different concentrations of Au, namely 100 g/ml, 500 g/ml and 1000 g/ml at a delay

PAGE 121

121 time of 20 s and gate width of 20 s. The spread of the LIBS signal response for these three different concentrations is shown below in Fig 5.5 0 1000 2000 3000 4000 5000 6000 7000 8000 266268270272274276 1000Au1 1000Au2 1000Au3 Wavelength nm Figure 5-4. Representative Au spectra showing range of LIBS response for concentration of 1000 g/ml Figure 5-5. Histogram of P/B for Au at 267.59 nm for concentrations of 100, 500 and 1000 g/ml

PAGE 122

122 10 20 30 40 50 60 70 80 90 020040060080010001200 y = 7.3533 + 0.075114x R= 0.99965 P/BConcentration m mm m g/ml Figure 5-6. Au Calibration curve for concentration range of 100-1000 g/ml at = 267.5 nm A linear calibration curve for Au, as shown in Fig 5.6 was obtained for concentration range of 100-1000 g/ml. The detection limit was determined using 3-s criterion, where the detection limit is defined as the analyte concentration which gives a signal response equal to three times of root mean square of noise level [85]. Using this definition and calibration curve, the detection limit for Au in current study was determined to be 29 g/ml. For calcium analyte, the spectra were collected at a delay of 30 s and gate width of 20 s. A series of spectra were recorded in sets of 100 for concentrations of 0.5, 1, 2.5 and 5 g/ml. A representative calcium spectrum is shown in Fig 5.7 for concentration of

PAGE 123

123 1 g/ml. The spectra were selected for calculating the average peak-to-base ratio using a two-step selection process. In the first step, ratio of P/B of two Calcium emission peaks at 393.37 nm (0-25 414 cm-1) and 396.85 nm (0-25 192 cm-1) was calculated and all the peaks with ratio less than 1.1 and greater than 2.4 were rejected, thus rejecting those peaks which could be saturated or self absorbed. In the second step, the spectra with P/B value less than 10 for 393.37 nm wavelength were rejected, as those contained essentially noise as discussed above. The remaining spectra were then averaged to obtain P/B value at 393.97 nm. The calibration curve for Ca is shown below in Fig 5.8. Using 3-s criterion as described above, the detection limit for Ca in current study was determined to be 0.05g/ml. This detection limit corresponds to an average of 50 droplets, each of 65 pl volume of Ca analyte, which corresponds to average mass of 163 fg while absolute mass of 3.25 fg in each droplet. 500 1000 1500 2000 2500 3000 3500 391392393394395396397398Intensity (a.u.)Wavelength, nm Figure 5-7. Representative Ca spectra 1g/ml

PAGE 124

124 20 40 60 80 100 120 140 160 180 012 3 4 5 6 y = 21.957 + 31.355x R= 0.9945 P/BConcentration, m mm m g/ml Figure 5-8. Ca calibration curve for concentration range of 0.5-5 g/ml Based on typical ICP detection limits for Ca, it is expected to be about three orders of magnitude better than Au In the current study, this ratio is about 600, which is in excellent agreement with the ICP detection limit values. Ca detection limits in various LIBS studies are listed here for isolated droplets and aqueous samples. Table 5-2. Calcium detection limits in various LIBS studies for isolated droplets and aqueous samples[85,87, 90,91,92]. Study Ca detection Limit Comments Current study 0.05 g/ml (50ppb) Isolated single droplet in LIBS Janzen et al. 20 ng/g (20ppb) Isolated single droplet in LIBS Lo et al. 3 ppb ( Ca I emission line) Aqueous sample using 193 nm ArF Knopp et al. 130 ppb Aqueous sample using non193 nm Archontaki et. al 0.4 ppm (400ppb) Isolated single droplet in LIBS Cremers et. al 0.8 ppm (800ppb) Liquid sample using spark

PAGE 125

125 It can be seen from the above table, the Ca detection limit obtained in this study is comparable or better than other LIBS studies on aqueous analyte or single droplets as the analyte. Archonataki et. al. noted that the difference in detection limit value as compared to Cremers et. al. was due to differences in the breakdown mechanism in isolated droplet as compared to liquid media. The differences suggested that the breakdown in isolated single droplets occurs in air and not inside the droplet, while some previous studies have suggested otherwise. ICP-MS, and ICP-OES can achieve limits of detection in the ppt range. Even though the relative detection limit is reduced for the LIBS system as compared with ICP-MS, and ICP-OES systems, it has the advantages beyond the absolute limit of detection. Robustness and simplicity of the LIBS provide an easy means of sub-nanoliter analyte introduction, which can be useful in analytical conditions where low analyte volume is available while the analyte concentration is not the limiting factor. In such cases, detection of absolute mass becomes the key, and the LIBS system can provide excellent results, comparable to ICPMS and ICP-OES. Local Temperature Measurements Three analyte species were examined (Mg, Ca & Lu) with the goal of measuring the analyte electronic temperature (i.e. excitation temperature) evolution following injection of single droplets into the laser focal volume: Single 50 m droplets (~500-2500 g analyte/ml) were generated using the scientific grade droplet generator and injected into the plasma volume. Delay generators were used to control the droplet injection and laser pulse timing such that the droplets were centered on the laser focal volume when the laser was fired.

PAGE 126

126 LIBS Analysis: Two different Nd:YAG pulsed lasers, one operating at its fundamental wavelength (1064 nm, 5ns, 5 Hz), and other operating at 532 nm (5 ns, 5 Hz), were used. The 1064-nm laser was used for creating the laser-induced plasma, while the 532-nm laser was used for imaging. Plasma emission was collected on-axis with a pierced mirror and fiber-coupled to a 0.275-m spectrometer (0.12-nm resolution) and spectral data were recorded using an iCCD detector array. Data were recorded over delay times from 1 to 60 s. Representative spectra are shown below for the Mg Figure 5-9. Image of plasma emission showing 532-nm probe laser scattering from a single ~50 m Lu-doped droplet particle. (Fig 5-10) and Lu (Fig 5-11) analyte samples for different delay times. As shown, the ratio of the ion and neutral lines varies with time. Boltzmann plots are used to calculate the analyte electronic excitation temperature using Lu, with the relevant spectral lines

PAGE 127

127 0 1 1052 1053 1054 1055 105276278280282284286288Intensity (a.u.)Wavelength (nm) Figure 5-10. Representative spectra of Mg corresponding to delay times of 5 s (top), 15 (middle) s and 30 s (bottom). 0 1 1052 1053 1054 1055 1056 1057 105335 340 345 350 355 360Intensity (a.u.)Wavelength (nm) Figure 5-11. Representative spectrum of Lutetium corresponding to delay times of 1 s (top), 10 s (middle) and 30 s (bottom)

PAGE 128

128 and the spectroscopic data included in Table 5-3. The Einstein-Boltzmann equation for intensity of a spectral line, under LTE condition and for optically thin plasma, is given as ()hcE kT NAg Ie uTl-= (5-1) where N is atomic density, A is the transitional probability for emission; g is the statistical weight of the excited state, h is the Plancks constant (h= 6.6 x 10-34 J.s), c is the velocity of light (c= 3.0 x 108 m/s), E is the energy of the excited state, k is the Boltzmann constant (k = 1.38x 10-23J/K),T is the excitation temperature in K, u(T) is the partition function which represents sum of the weighted Boltzmann functions of all the discrete energy levels and is the wavelength of the emitted light. Considering the electronic transitions of the same element, the partition functions and the atomic densities will be constant. By rearranging and taking logarithms on both the sides, the following equation is obtained ln()()Ihc EC AgkT l =+ (5-2) By plotting the equation, with energy on the horizontal axis and ln() I Ag l on the vertical axis, the excitation temperature can be measured from the slope of the resulting line. The Lu lines at 347.25, 350.74 and 355.44 nm were used for calculating the excitation temperature using the Boltzmann plot. As can be seen from the Table 5-3 that these Lu spectral lines provide a wide spread in their upper energy levels which are very important for accurate temperature measurements using the Boltzmann method. A representative Boltzmann plot is shown Fig 5-12at 40 and 60s of delay times.

PAGE 129

129 Table 5-3. Lu spectroscopic data [73] Wavelength, nm Energy, cm 1 Ag s 1 339.71 11,796-3.5 x 10 8 347.25 12,435-4.2 x 10 8 350.74 0-28,503 0.2 x 10 8 355.44 17,332-14.0 x 10 8 -8 -7 -6 -5 -4 -3 -2 -1 2.5 1063 1063.5 1064 1064.5 1065 106 y = 7.7734 3.2558e-06x R= 0.9994 Ln( l l l l *I/Ag)Energy, cm-1 Figure 5-12. Boltzmann plots for Lu II lines at 347.3 nm, 350.7 nm and 355.7 nm respectively. Upper energy levels for these lines are 41225, 28503, 45458 cm-1 respectively. Recent work by Diwakar et al. has shown that plasma-particle interactions are confined to localized regions about the particles at early to moderate delay times. While the mass of the particle is generally insufficient to perturb the average plasma properties, it is not unreasonable to expect changes in T and ne as well as ionization fractions, around the particle vicinity. Such changes can directly affect the analyte atomic emission, for example, by changing the ion to neutral fraction, the electronic

PAGE 130

130 temperature, or the upper state population. With increased time, diffusion of heat and mass may mitigate such effects. In a numerical simulation of plasma-particle interactions, Dalyander et al..calculated the Lewis number to be ~1, which suggests 3000 4000 5000 6000 7000 8000 9000 10000 010203040506070Excitation Temperature (K)Delay Time ( m mm m s) Figure 5-13. Temporal temperature evolution of the analyte by introducing single droplet of Lu at 1 Hz repetition rate. The temperature was calculated by using Boltzmann plots for Lu II lines at 347.3 nm, 350.7 nm and 355.7 nm respectively. Upper energy levels for these lines are 41225, 28503, 45458 cm1 respectively. comparable time scales for heat and mass transfer. Recall that the Lewis number () Le D a = is a non-dimensional ratio of the thermal diffusivity, and mass diffusivity, D, which is indicative of relative roles of heat and mass transfer. This is consistent with the above discussion regarding localized interactions between the analyte and bulk plasma.

PAGE 131

131 As shown in Figures 5-13 and 5-14, the temporal temperature evolution can be divided in two regions: region of local temperature perturbation where the analyte absorbs energy leading to decrease in Tlocal, and a region of plasma analyte equilibrium where Tlocal starts approaching Tbulk. Fig 5-15 shows the temperature history of Si atoms in an ICP, where T is initially suppressed and then approaches TICP as heat diffuses inward. This is analogous to the droplet behavior in the laser-induced plasma. Figure 5-14. Ion-to-Neutral ratio of Mg lines at 280.27 nm and 285.21 nm respectively. Mg is mixed with Ca in ratio of 1:5 in this experiment and overall conc. of 3,000 g/ml.

PAGE 132

132 Figure 5-15. Silicon atom temperature plotted against residence time by introducing single droplet of Si analyte in a ICP-MS system at 1 Hz repetition rate. Temperature is calculated using Boltzmann plots for Si lines at 212 nm, 251 nm, and 288 nm[86]. Diffusion Coefficient Measurements The binary diffusion coefficient can be calculated using the ChapmanEnskog theory, which gives the diffusion coefficient as [93] 1/2 2(4/) 3 16(/)B AB AB D uABDKTMW D f PRTp ps= W (5-3) Here, A represents nitrogen, while B is represents either calcium or hydrogen, B K denotes the Boltzmann constant, AB MW is the harmonic mean of molecular weights of the two species A and B, P and T are Pressure and Temperature, respectively, u R denotes the Universal Gas constant, AB s is the arithmetic mean of the hard sphere collision diameters of species A and B, D f is the theoretical correction factor which here has been assumed to be unity. D W is the dimensionless parameter given by the following expression ** (.)(.)(.) ()D B DTFTHT ACEG T eeeW=+++ (5-4)

PAGE 133

133 where the values of constants are given as A= 1.06036, B = 0.15610, C = 0.19300, D = 0.47635, E = 1.03587, F = 1.52996, G = 1.76474, H = =3.89411 [91]. Finally, T is a dimensionless Temperature parameter given as, 1/2 () B AB KT Tee= (5-5) Where is Leonard-Jones energy for different species. Equation (5-3) can be reduced to 3 2 3 2 11 1.86.10 AB AB ABDT MM D Ps-+ = W (5-6) where T is in K, P is in atm units, AB s is in. Based on above given expression, the theoretical diffusion coefficient for each species can be calculated. In the current study, calcium and hydrogen diffusion has been studied. The species specific values have been listed in Table 5-4. Based on these parameters, 10 NH NCaD D @ thus the hydrogen diffusion is estimated to be about one order of magnitude faster than calcium Diffusion coefficient was calculated for the temperature range of 3000-25,000K. The diffusion coefficient values ranged from 0.0005-0.0243 m2/s for calcium in air and from 0.0036-0.1223 m2/s for hydrogen in air. Table 5-4. Calcium, Hydrogen and Nitrogen properties for calculation of diffusion coefficients [94,95] Properties Calcium Hydrogen Air Lennard Jones energy, /KB (K) 4850 59.7 78.6 Hard sphere collision diameter, 4.063 2.827 3.711 Molecular Weight MW 40.08 2.01 28.97

PAGE 134

134 These values are lower than experimentally measured values for calcium diffusing in air in previous studies [18]. This is because the above discussed ChapmanEnskog theory did not take into account various other species, especially the ions, participating in the diffusion process. Also the hydrogen properties used in the calculations are for molecular hydrogen, which is not the case in the current study, as the hydrogen in the plasma is originated from the water droplet and hence the LeonardJones properties will be different for the hydrogen atom attached to water molecule as compared to the hydrogen molecule. In order to get the better estimate of the diffusion coefficients, calculations were done taking into account the properties of ionized calcium and hydrogen attached with water molecule. Since the Lennard-Jones properties for Ca+1 ion were not available, the parameters of Ca and Ca+2 were averaged for the calculations. The diffusion coefficient values ranged from 0.001-0.0352 m2/s for calcium in air and from 0.0094-0.3212 m2/s for hydrogen in air. This value is closer to the experimental value observed in previous study for diffusion of calcium in air (0.0170.063 m2/s) [18]. The ratio of diffusion coefficients for hydrogen and calcium still remains ~10. Table 5-5. Lennard-Jones parameters for Ca ion and H atom attached to water [96,97] Atom n, J/mole Ca +1 418.45 3.465 H of water 192.28 0.400 In this study, same setup as described above in Fig 5-2. is used. For imaging, a hydrogen-alpha line filter at wavelength of 656.28 nm (FWHM 1.5 nm) was used to image hydrogen emission from the analyte droplet in the LIBS plasma at different times. Fig 5-16 shows the hydrogen emission images for different times after the plasma

PAGE 135

135 inception (t=zero). It can be seen that the hydrogen emission is visible as early as 250 ns after the. plasma is formed, and within 1 s, the hydrogen has diffused throughout most Figure 5-16. Spectral images of hydrogen diffusion measured temporally. Hydrogenalpha line filter (656.28 nm) was used with line width of 10 nm. The bar represents 1 mm. 0.4 0.6 0.8 1 1.2 1.4 20040060080010001200Hydrogen alpha emission burst diameter, mm Delay time, ns Figure 5-17. Average diameter of Hydrogen emission burst as a function of delay time. Error bars depict standard deviation, where N= 50.

PAGE 136

136 of the plasma volume. The size of hydrogen emission cloud was calculated directly from the images. Fig 5.17 shows the average hydrogen emission cloud diameter plotted as a function of the delay time. Table 5-6. Hydrogen Diffusion coefficient calculations Delay, ns Hydrogen cloud size mm Stdev m 2 /sec (Two-point Slope) 250 0.470084 0.177764 0.883916 500 0.683286 0.247469 0.983601 750 0.906794 0.302027 1.421587 1000 1.238286 0.415923 2.844302 Average diffusion coefficient was estimated to be 1.5 m2/s, using a two-point slope method which is slightly higher than the value obtained from theoretical estimation. Ca diffusion coefficient estimated from a previous study was reported to be 0.04 m2/s [18]. The ratio of hydrogen-to-Calcium diffusion coefficient is ~40, which is about four times larger as predicted by the theoretical model. However, the overall agreement is considered quite good, given the possible variations in dissociation rates between hydrogen and calcium, hence differences in species gradients. The hydrogen diffusion coefficient measured in this study, therefore still provides a good estimate of the diffusion process taking place inside the plasma at very early times in the plasma. Fig 5.18 shows the calcium diffusion images as a function of delay time. It can be seen that the calcium emission is not evident at very early times in the plasma (i.e first few 100 ns). Calcium emission is visible only after 1.25 s, and it becomes prominent after about 2-3 s. In order to corroborate the calcium diffusion images as shown in Fig 5.18, calcium spectra was measured as a function of time. The aim of this exercise was to see if any prominent changes in the LIBS spectral response was present as a function of time. The

PAGE 137

137 P/B of the Ca spectra was calculated at 393.37 nm and is shown as a function of time in Fig 5-19. It can be seen in the figure that the slope of the curve changes first at 1 s, and Figure 5-18. Spectral images of Calcium diffusion measured temporally. Calcium line filter (396 nm) was used with line width of 3 nm. The bar represents 1 mm. then at about 3s, which is consistent with the calcium images as seen in Fig 5.18. This shows that in the LIBS plasma, the droplet first desolvates losing primarily water, with the water rapidly dissociating to hydrogen, which rapidly diffuses within the plasma. The Ca analyte starts significantly diffusing only after 2-3 s, which supports previous studies about finite time scales for heat and mass transfers and strengthens the localized perturbation argument. These results are in good agreement with analogous droplet measurements in an ICP, which show hydrogen emission prior to calcium emission. Together, these studies suggest that the water is primarily evaporated leaving the analyte behind for the most part.

PAGE 138

138 0 40 80 120 160 0123456P/BDelay Time, m mm ms Figure 5-19. LIBS response for Ca as a function of delay time This process appears to take about 2-3 s in the :LIBS plasma for the current 50-m droplets. Conclusions One of the key achievements of this study was the successful introduction of analyte species as single isolated droplet in the LIBS plasma. Low sample introduction in sub-nanoliter to picoliter range was achieved which is useful in the low sample availability conditions. Excellent absolute detection limits for gold and calcium were achieved. Another major accomplishment of this study was the measurement of local excitation temperature and diffusion coefficients. The results showed the perturbation of localized properties in the vicinity of the analyte species at early times in the plasma which is consistent with the fractionation and matrix effect studies as discussed in the previous chapter. Some of the key findings from this study are as follows: Sample introduction in the sub-nanoliter range was achieved using scientific grade microdispenser, with successful LIBS detection.

PAGE 139

139 Calibration curves for Ca and Au were obtained and detection limits were calculated using the 3-s criterion. The detection limits were 0.05 and 29 g/ml for Ca and Au, respectively. Average absolute detection limit for Ca was determined to be 3.25 fg. The temporal temperature evolution can be divided in two regions: a region of local temperature perturbation where the analyte absorbs energy leading to a decrease in Tlocal, and a region of plasma analyte equilibrium where Tlocal starts approaching Tbulk. In order to estimate the diffusion coefficient of hydrogen, spectral imaging experiments were performed temporally using the ICCD camera and H and calcium emission line filters. Results show extremely fast diffusion rates for hydrogen as compared to calcium and strengthens the finite time scales for heat and mass transfer arguments for the heavier analyte species.

PAGE 140

140 CHAPTER 6 CONCLUSION AND FUTURE WORK Laserinduced breakdown spectroscopy is a relatively new analytical technique as compared to well established techniques including ICP-MS and ICP-OES. Even though LIBS is new in the analytical field, it offers ample advantages in certain aspects, which makes it an upcoming analytical method to consider in the next decade. Simplicity, robustness, in situ and multi-elements measurement capabilities, and portability make it a very attractive technique for certain applications. Along with all its advantages, LIBS comes with certain drawbacks which include low sensitivity, matrix and fractionation effects, and calibration related issues These may all apply when LIBS is used to study aerosols. In order to make LIBS an effective quantitative technique, these drawbacks need to be understood and hopefully overcome. LIBS certainly has the capacity to complement other well established techniques in the world of analytical diagnostic methods. For any new scientific technique to develop fully, understanding of the fundamentals of the processes and mechanisms taking place is very essential. With the advent of any new technique, two types of research may be pursued in the scientific community: those that focus on the applicability of the new scientific tool to various situations, and others that focus on understanding the fundamentals and physics behind various nuances of the technique. Both research directions lead to development of the new scientific technique, but it is the latter that can usher any technology into its next stage of development. In this study, the main focus has been to understand the fundamentals of various aspects of LIBS and to provide answers to various questions in order to overcome the above mentioned drawbacks. Since 1963, when the first

PAGE 141

141 breakdown study was reported, to 1983, when the first LIBS experiments were reported, the technique has come a long way, but the majority of fundamental understanding of the processes that occur has taken place in last few years, which has propelled LIBS in the direction of being a well established analytical technique. This study, which mostly focused on LIBS involving aerosols, has been able to unravel some of the mysteries and provide knowledge that will be valuable to LIBS community as a whole. The plasma forms the essence of a LIBS methodology, hence it becomes the most important entity to study and investigate in order to understand the processes and mechanisms leading to a LIBS signal. LIBS processes can be broken down to three basic steps, namely, plasma formation, anayte introduction, and plasma-anlyte interactions. In this study, these three steps have been investigated in a laser-induced plasma, focusing mainly on the plasma particle interactions. Some very fascinating results have been obtained in these studies and are summarized below. Plasma Inception Study As mentioned before, plasma formation is core to the LIBS process and its understanding is important as it affects the plasma chemistry and plasma dynamics, which in turn affects the LIBS response. Temporal imaging of the plasma was done in different gases in order to understand how the very first breakdown occurred in different gases and how the plasma growth took place. As discussed in chapters 1 and chapter 4, multiphoton ionization and cascade growth processes are the most important processes taking place during the plasma inception, which provide the first free electrons for the plasma formation. This study showed that the plasma formation is initiated at various locations in the form of spots (i.e. individual breakdown points) on the optical axis ahead of the focal plane towards the laser direction. This phenomenon was observed in

PAGE 142

142 all three gases, although it differed in intensity, location, and number of spots. A plausible explanation for these multiple plasma initiation points is spherical aberration of the optical system, while the difference in the intensity and number of the plasma initiation points can be attributed to the difference in the breakdown threshold of the gases. Nitrogen was shown to have the most number of plasma initiation locations, i.e. spots, which were well separated and had the highest intensity as compared to helium and argon. Self-focusing caused by nitrogen molecules provides a plausible explanation for this difference in behavior as compared to the other two gases. The first ionization potentials are 14.35, 15.75 and 24.58 ev for nitrogen, argon and helium, respectively, which shows that the breakdown in nitrogen is easier as compared with helium and argon, which is an important factor in the breakdown process Also, it was observed that the plasma formation process was very repeatable with respect to time, implying that there was very little deviation in the plasma inception time that can be attributed to the high energy laser pulse (400 mJ per pulse). Also the experiments showed that with time, these multiple plasma initiation points coalesced to form a larger and more stable plasma. Further studies need to be conducted to understand the statistics of the plasma inception process and reasons for the variations amongst the different gases. This analysis is planned for future work Electron Density Measurements In this study, electron densities at very early times in the plasma were measured using a combination of Thomson scattering and Stark broadening methods. As r esearchers begin to focus more attention on LIBS using ultra-short laser pulses and very low pulse energies, the time-scales for spectral analysis become compressed, adding additional importance to the establishment of LTE conditions as they relate to analytical

PAGE 143

143 treatment of the plasma emission. It is readily assumed in the LIBS community, that the LTE is achieved at early times in the plasma when the electron density is well above 1017 cm-3. In this study, using the Stark broadening technique, the electron density was measured on order of 1019 cm -3 for the first 100 ns, and so it would generally be assumed that during this temporal region LTE was achieved. But Thomson scattering experiments revealed otherwise, specifically that LTE was not established for the first 40-50 ns after the plasma formation, which was evident by intense Thomson scattering caused by non-equilibrium conditions in the plasma (i.e. Te fTi). The results are consistent with the fact that the McWhirter criterion for LTE is a necessary but not sufficient condition, and should be used with caution. Hence LTE conditions are not i nitially realized despite the very high initial plasma electron densities, which suggest a temporal time scale of some tens of nanoseconds for establishment of equilibrium conditions following laser-induced breakdown. This suggests the time scale for the energy to cascade from the free electrons to ions and neutrals, presumably via collisional energy transfer. Such deviations from LTE as reported in the current study suggest the need for further scrutiny of the earliest temporal regimes of laser-induced plasmas to gain a better understanding of the energy transfer dynamics. Important outcomes of this s tudy are listed below: Electron density was measured as early as 5 ns delay in the plasma. High electron densities ,~ 1019 cm-3, were observed at early times. An opaque plasma for delay times of 10-20 ns, using a 532-nm probe. Correlation was observed between high electron density measurements and Thomson scattering observations. An optically thin plasma at longer delays with a 532-nm probe.

PAGE 144

144 A plausible explanation for the Thomson scattering is ion resonance peaks due to non-LTE conditions (Te fTi) at early times in plasma (~ 50 ns) Fractionation and Matrix Effects Study As mentioned above, plasma-particle interactions are the essence to the LIBS response, and the physics and mechanisms involved need to be understood very carefully. Precise calibration of the analyte spectral signal with respect to analyte concentration within a laser-induced plasma is necessary for achieving accurate results with the LIBS technique. A key assumption for LIBS is that complete dissociation of the constituent species within the laser-induced plasma results in independence of the analyte atomic emission signal on the analyte source & independence on the presence of other elements. But the critical question here is if it is valid all the time, or if there are any limitations in this assumption, or if there are any localized effects in the plasma which can affect the LIBS signal. The goal of this study was to answer these very important questions which can have serious consequences on understanding and calibrating the LIBS response. The fate of an analyte species in the plasma depends on various processes: vaporization, dissociation, atomization and diffusion. A discrete particle in the plasma undergoes all of these processes, which ultimately results in the atomic emission or LIBS signal. It is important to investigate the time scales of these processes so as to understand if the analyte species is instantaneously dissociated in the LIBS plasma, or if dissociation occurs in a finite time scale and that has any implications on the LIBS response. Considering the fate of the analyte in the plasma, it can follow two paths depending on the timescales of above mentioned processes. First scenario can be envisioned as an ideal case where the atomization and vaporization process, as well as heat and mass diffusion, takes place instantaneously, resulting in uniform plasma-

PAGE 145

145 analyte interaction with no localized perturbations. Such a scenario, or ideal case, will lead to a LIBS response which is directly proportional to the analyte concentration and will be independent of the original particle size and composition. Also the bulk plasma properties will be unperturbed due to the presence of the particles. But as suggested, this would be the ideal case, while previous studies by Hahn and co-workers, as well as the current study have shown the localized presence of the analyte species inside the plasma for 10-15 s after plasma formation, which suggests that the real scenario of plasma particle interactions might be different from the ideal case [18,57]. An alternative scenario for plasma-particle interactions can be envisioned where in the atomization, vaporization and heat and mass transfer processes are not instantaneous but rather limited by diffusion and dissociation time scales, resulting in a localized analyte in the plasma where the localized conditions can be different from the bulk plasma conditions. In such a scenario, addition of other elements can perturb the localized conditions, thereby altering the LIBS response. The present study examined the phenomena of analyte dissociation and analyte emission response for aerosol particles comprised of either one or two elements, and the resulting effects on analyte calibration response. Three primary analytes were examined, namely Na, Cd, and Mg, with the goal of providing different matrices of the analyte within a range of submicron-sized aerosolized particulates. Pure Na, Cd or Mg particles were prepared from the nebulization of ICP-grade solutions, while binary particles of Na, Cd or Mg were prepared by the addition of Zn, Cu or W. It was observed that the presence of additional mass has the general result of enhanced analyte response, while the plasma continuum emission remains constant. Na atomic emission was enhanced by

PAGE 146

146 28-50% with the addition of mass, while Mg and Cd exhibited similar trends. At long delay times (30 to 60 s), the enhancement factors decay back toward unity, which was attributed to complete dissociation and diffusion of the analyte atoms throughout the plasma at longer delays. The Mg data showed a 10% increase in neutral/ion ratio at a delay time of about 5 s, while the Cd data did not display such a clear trend. Previous studies have demonstrated the plasma robustness with regard to aerosol particle sampling. However, more recent work, and the current study have shown that plasmaparticle interactions are confined to localized regions about the particles [Hohreiter et. al.]. While the mass of submicron-sized particles is generally insufficient to perturb average plasma properties, it is not unreasonable to expect changes in T, ne and ionization fractions around the particle, which then directly affect the analyte emission. With increased time, diffusion of heat and mass mitigate such effects. Localized temperature measurements provided evidence for matrix effects in the LIBS plasma as the temperature was perturbed by additional particle mass in the vicinity of the analyte, hence affecting the LIBS signal. Temporal evolution of the localized temperature showed suppression of the temperature during early times, which can be attributed to loss of plasma energy during vaporization of the analyte. At longer times, the local temperature increases due to heat transfer from the surrounding plasma. The overall message from this study was that the dissociation and diffusion time scales in LIBS plasma are finite, so important care should be taken in quantitative LIBS analysis, taking into consideration the temporal plasma evolution, localized perturbations, and finite time scales for the dissociation and diffusion processes. Key observations from this study are as follows:

PAGE 147

147 Added mass within an aerosol particle can perturb local T and electron density, altering the emission signal. Ionization fractions can change via Saha equation due to local perturbations. Atomic emission can change via Boltzmann populations and ionization fractions. Rates of mass and heat diffusion play key roles, with temporal considerations being important. Investigation of the Single Droplet in LIBS-Plasma The goal of this study was to understand the plasma particle interactions by the introduction of single droplets containing an extremely small volume of the analyte sample. Analyzing single droplets and particles in analytical plasmas can provide great insight into the various fundamental processes namely vaporization, desolvation, atomization, plasma-particle interaction, and so on. Single droplet study is important from a fundamental as well as an application point of view. Measurements of localized temperatures in the vicinity of the analyte, and the diffusion time scales for hydrogen and calcium were important for understanding the fundamental processes in the LIBS plasma, while developing a sub-nanoliter sample introduction system in the LIBS plasma was important from an application point of view. Picoliter range analyte introduction in the plasma can have an advantageous effect as the plasma cooling will be limited to localized region as compared to the analyte introduction using nebulizers, where heavy mass loading can lead to overall plasma cooling and thereby interfere with the LIBS signal. In this study, a sub-nanoliter sample introduction system was developed with excellent absolute limit of detection. The Ca detection limit obtained in this study was comparable or better than other LIBS studies on aqueous and single droplet analyte. Detection limits of 0.05g/ml and 29 g/ml were determined for Ca and Au using single droplet sample introduction in the LIBS plasma. This detection limit corresponds to an average of about 50 droplets, each of 65 pl volume of Ca analyte, corresponds to an

PAGE 148

148 absolute mass of 3.25 fg ,which is excellent for analytical studies where absolute mass detection is more critical but the availability of the analyte sample is limited. Despite this success, the relative detection limit is low for a LIBS system compared with ICPMS and ICP-OES systems. However, LIBS has the advantage over absolute limit of detection. Robustness and simplicity of LIBS provided an easy way of sub-nanoliter analyte introduction, which can be useful in analytical conditions where low analyte volume is available while the analyte concentration is not the limiting factor. In such cases, detection of absolute mass becomes the key, and LIBS system can provide excellent results, comparable to ICP-MS and ICP-OES. Three analyte species were examined (Mg, Ca & Lu) with the goal of measuring the analyte electronic temperature evolution following injection of single droplets into the laser focal volume. The temporal temperature evolution obtained in this study showed two regions: a region of local temperature perturbation where the analyte absorbs energy leading to decrease in Tlocal, and a region of plasma analyte equilibrium where Tlocal starts approaching Tbulk. This study corroborates previous studies and confirms the perturbation of localized plasma conditions at earlier times in plasma evolution. As discussed above, it was suggested that the finite time scales of heat and mass transfer play an important role in localized plasma perturbations in the vicinity of the analyte, and which in turn affect the LIBS signal. In this study, the goal was to measure the diffusion coefficients of the analyte by imaging the plasma-analyte interactions temporally and study its affect on the LIBS signal. A hydrogen-alpha line filter at the wavelength of 656.28 nm (FWHM 10 nm) was used to image hydrogen emission from

PAGE 149

149 the analyte droplet in the LIBS plasma at different times. The average diffusion coefficient for hydrogen was estimated to be 1.5 m2/s, while Ca diffusion coefficient, as estimated from previous study, was reported to be 0.04 m2/s. The results also showed that the Ca emission is delayed as compared to hydrogen diffusion. Calcium emission was not evident at very early times in the plasma becoming visible only after about 1.25s, and more prominent after about 2-3 s. This shows that in the LIBS plasma, the droplet first dissolvates, and within about one microsecond, hydrogen diffuses rapidly inside the plasma. Ca analyte starts diffusing only after 2-3 s, presumably after complete dissolvation, which supports previous studies and arguments by about finite time scales for heat and mass transfers, and strengthens the localized perturbation argument. Key observations from the single droplet study are listed as follows: Sample introduction in the sub-nanoliter range was achieved using a scientific grade microdispenser. Calibration curves for Ca and Au were obtained and detection limits were calculated using the 3-s criterion. The detection limits were 0.05 and 29 g/ml for Ca and Au, respectively. Average absolute detection limit in 50 droplets of 65 pl volume of Ca analyte was determined to be 3.25 fg per droplet. The temporal temperature evolution can be divided into two regions: a region of local temperature perturbation where the analyte particle or droplet absorbs energy leading to decrease in Tlocal, and a region of plasma analyte equilibrium where Tlocal starts approaching Tbulk. In order to estimate the diffusion coefficient of hydrogen, spectral imaging experiments were performed temporally using an ICCD camera and the H and calcium emission line filters. Results show extremely fast diffusion rates for hydrogen as compared to calcium, which strengthens the arguments for finite time scales for heat and mass transfer. Aerosol Sampling Study For individual particle measurements with the LIBS technique, understanding of the sampling statistics is critical to obtain a meaningful quantitative measure of the

PAGE 150

150 sample analyte. The current results demonstrate the statistical nature of LIBS-based aerosol sampling and provides corroboration of Poisson-based models to describe the sampling problem. Sampling of aerosols with LIBS is affected by low signal-to-noise ratios if traditional ensemble averaging is applied. An appropriate choice of a conditional analysis algorithm can therefore enhance the signal-to-noise ratio, with the use of Poisson statistics being a useful analysis tool. Experiments were setup to sample ambient air during a dust storm event, and to see if the LIBS instrument was able to monitor or detect any increase in signal due to particulates in the ambient air. Key observations from this study are as follows: Significant calcium-rich particle hits were recorded, although no significant transient was observed with the passing of the dust storm. Poisson statistics provided accurate sampling rate predictions. Poisson statistics enable analysis & prediction of key regimes for appropriate sampling algorithms. In summary, the results from this study provide additional insight into the plasma formation processes and the role of plasma-particle interactions pursuant to quantitative aerosol analysis with laser-induced breakdown spectroscopy. Based on current study and other recent studies by our group and others, the current picture that is emerging concerns a complex interaction between the plasma and the aerosol particle, during which the finite time-scales of particle dissociation, and heat and mass transfer are fundamental processes. These results are important not only for the LIBS community but also for the other communities including ICP-MS and ICP-OES, as plasmas form an essential component of these analytical methods as well. Ultimately, it is believed that a better understanding of the fundamental plasma processes is key in order to improve and develop LIBS as an accepted analytical method.

PAGE 151

151 Future Work Plasma inception study: It will be interesting to carry out further experiments to understand the breakdown process in different gases and understand the plasma inception process which has implications on the plasma dynamics. It is suggested as future work to study the statistics of plasma initiation with respect to laser energy deposition and spatial location. In the current study multiple plasma initiations were observed and these breakdown images can be used to quantify the energy depositions and spatial deviations. Figure 6.1 shows this approach graphically. Figure 6-1. Quantification of plasma initiation processes. An estimate of the area under the curve, FWHM, spatial locations can give a.plethora of information regarding the breakdown mechanisms and statistical nature. Thomson scattering experiment: Another proposed future work in the continuation to current study will be to measure the electronic satellite peaks in the

PAGE 152

152 Thomson scattering at very early times in the plasma and thereby evaluate electron density and temperature via Thomson scattering. In the current study, the satellite peaks were not able to be observed. More elaborate experimental setup will be useful in measuring both electron as well as ion peaks in Thomson scattering, and the temporal evolution of these peaks can give very valuable information regarding early plasma dynamics and the establishment of LTE conditions. Single droplet experiment: More future experiments are proposed using the single droplet analyte introduction method. In the current study, the size of droplets was about 50 m. It is suggested to develop the droplet introduction system so as to get smaller sized droplets which will result in more precise localized temperature and electron density measurements. Increasing the distance between the plasma and the droplet generator can be one of the ways to provide enough time for the droplet to sufficiently desolvate. Droplets can be allowed to pass through a heated tube to increase the rate of desolvation. Also it is suggested to carry out more plasma emission spectral imaging experiments to estimate the diffusion coefficients of various analyte species including calcium and gold which were used in the current study.

PAGE 153

153 REFERENCES [1] J. D. Winefordner, I. B. Gornushkin, T. Corell, E. Gibb. B.W. Smith, N. Omenetto. Comparing several atomic spectrometric methods to the super stars: special emphasis on laser induced breakdown spectrometry, LIBS, a future super star, J. Anal. At. Spectrom. 19 ( 2004) 1061-1083. [2] L.J. Radziemski. From laser to LIBS, the path of technology development, Spectrochim. Acta Part B 57 (2002) 1109-1113. [3] S. Yalcin, D. R. Crosley, G. P. Smith, G.W. Faris, Spectroscopic characterization of laser-produced plasma for in situ toxic metal monitoring, Haz. Waste and Haz. Materials 13 (1996) 51-61. [4] B.T. Fisher, H.A. Johnsen, S.G. Buckley, and D.W. Hahn Temporal gating for the optimization of laser-induced breakdown spectroscopy detection and analysis of toxic metals. Applied Spectroscopy, 55 (2001)1312-1319. [5] S.G. Buckley, H.A. Johnsen, K.R. Hencken, and D.W. Hahn. Implementation of laser-induced breakdown spectroscopy as a continuous emissions monitor for toxic metals, Waste Management, 20 (2000) 455-462. [6] J.E. Carranza, B.T. Fisher, G.D. Yoder, and D.W. Hahn. On-line analysis of ambient air particles using laser-induced breakdown spectroscopy, Spectrochimica Acta Part B, 56 (2001) 851-864. [7] R. Noll R, H. Bette A. Brysch, M. Kraushaar I. Mnch L. Peter L,V. Sturm. Laser-induced breakdown spectrometry applications for production control and quality assurance in steel industry. Spectrochimica Acta Part B 56 (2001) 637649 [8] A. J. Ball, V. Hohreiter, D. W. Hahn. Hydrogen Leak Detection Using LaserInduced Breakdown Spectroscopy, Applied Spectroscopy, 59 (2005) 348-353. [9] S. Palanco, C.L. Moreno, J.J. Laserna, Design, construction and assessment of a field-deployable laser-induced breakdown spectrometer for remote elemental sensing. Spectrochimica Acta Part B, 61 (2006) 88-95 [10] A. Giakoumaki, K. Melessanaki and D. Anglos, Laser-induced breakdown spectroscopy (LIBS) in archaeological scienceapplications and prospects, Analytical and Bioanalytical Chemistry, 387 (2007) 749-760.. [11] P.B. Dixon and D.W. Hahn. On the Feasibility of Detection and Identification of Individual Bioaerosols Using Laser-Induced Breakdown Spectroscopy, Analytical Chemistry, 77 (2005)631-638

PAGE 154

154 [12] J. Diedrich and S. J. Rehse, Escherichia coli identification and strain discrimination using nanosecond laser-induced breakdown spectroscopy, Appl. Phys. Letter, 90 (2007)163901. [13] A. K. Knight, N. L. Scherbarth, ,D. A. Cremers, M.J.Ferris, Characterization of Laser-Induced Breakdown Spectroscopy (LIBS) for Application to Space Exploration, Applied Spectroscopy, 54 (2000) 331-459. [14] F. Colao, R. Fantoni, V. Lazic, A. Paolini, F. Fabbri, G. G. Ori, L. Marinangeli, A. Baliva. Investigation of LIBS feasibility for in situ planetary exploration: An analysis on Martian rock analogues, Planetary and Space Science, 52 (2004) 117-123. [15] B. Hettinger, V. Hohreiter, M. Swingle, and D.W. Hahn, Laser-Induced Breakdown Spectroscopy for Ambient Air Particulate Monitoring: Correlation of Total and Speciated Aerosol Counts, Applied Spectroscopy, 60 (2006) 237-245 [16] A. W. Miziolek, V. Palleschi, I. Schechter, (eds.) Laser-Induced Breakdown Spectroscopy: Fundamentals and Applications, Chapter 3, From signal to LIBS: A complex route to quantitative analysis, E.Tognoni, V. Palleschi, M. Corsi, G. Cristoforreti, N. Omenetto, I. Gornushkin, B.W. Smith, J. D. Wineforder, Cambridge press (2006). [17] B.T. Fisher, H.A. Johnsen, S.G. Buckley, and D.W. Hahn Temporal gating for the optimization of laser-induced breakdown spectroscopy detection and analysis of toxic metals. Applied Spectroscopy, 55 (2001) 1312-1319. [18] V. Hohreiter and D.W. Hahn, Plasma-Particle Interactions in a Laser-Induced Plasma: Implications for Laser-induced Breakdown Spectroscopy Analytical Chemistry, 78 (2006) 1509-1514. [19] V. Hohreiter, J.E. Carranza, and D.W. Hahn, Temporal analysis of laserinduced plasma properties as related to laser-induced breakdown spectroscopy Spectrochimica Acta Part B, 59 (2004) 327-333. [20] C.G. Parriger D.H. Plemmons, E. Oks, Balmer series H measurements in a laser-induced hydrogen plasma, App. Opt. 42 (2003) 5992-6000 [21] M. Villagran-Muni, h. Sobral,E. Camps, Shadowgraphy and interferometry using a CW laser and a CCD of a laser-induced plasma in atmospheric air, IEEE Transactions on Plasma Science 29(4) (2001) 613-616. [22] Z. Whang, B. Hattendorf, D. Gunther. Analyte response in laser ablation inductively coupled plasma mass spectrometry, J. Am. Soc. Mass Spectrom. 17 (2006) 641-651.

PAGE 155

155 [23] M. Guillong, D. Gunther. Effect of particle size distribution on ICP-induced elemental fractionation in laser ablation-inductively coupled plasma-mass spectrometry, J. Anal. At. Spectrom. 17 (2002) 831-837. [24] C. Chaleard, P. Mauchien, N. Andre, J. Uebbing, J. L. Lacour, C. Geertsen. Correlation of matrix effects in quantitative elemental analysis with laser ablation optical emission spectrometry, J. Anal. At. Spectrom. 12 (1997) 183188. [25] J. M. Mermet. Revisitation of the matrix effects in inductively coupled plasma atomic emission spectrometry: the key role of the spray chamber, J. Anal. At. Spectrom. 13 (1998) 419-422. [26] D. Gunther. Laser-ablation inductively-coupled plasma mass spectrometry, Anal. Bioanal. Chem. 372 (2002) 31-32. [27] J. Koch, H. Lindner, A. V. Bohlen, R. Heregenroder, K. Neimax. Elemental fractionation of dielectric aerosols produced by near-infrared femtosecond laser ablation of silicate glasses, J. Anal. At. Spectrom. 20 (2005) 901-906. [28] J. Koch, A. V. Bohlen, R. Heregenroder, K. Neimax. Particle size distributions and compositions of aerosols produced by near-IR femtoand nanosecond laser ablation of brass, J. Anal. At. Spectrom. 19 (2004) 267-272. [29] G. Rezaaiyaan, J.W. Olesik, G.M. Hieftje. Interferences in low-flow, low-power inductively coupled plasma, Spectrochim. Acta Part B 40 (1985) 73-83. [30] K. OHanlon, L. Ebdon, M. Foulkes. Effect of easily ionisable elements on the mass transport efficiency of solutions and slurries used in plasma emission spectrometry, J. Anal. At. Spectrom. 12 (1997) 329-331. [31] R. Russo, X. Mao, J. J. Gonzalez, S. S. Mao. Femtosecond laser ablation ICPMS, J. Anal. At. Spectrom. 17 (2002) 1072-1075. [32] R.E. Russo, X.L. Mao, O.V. Borisov, H.C. Liu. Influence of wavelength on fractionation in laser ablation ICP-MS, J. Anal. At. Spectrom. 15 (2000) 11151120. [33] D. Bleiner and D. Gunther. Theoretical description and experimental observation of aerosol transport processes in laser ablation inductively coupled plasma mass spectrometry, J. Anal. At. Spectrom. 16 (2001) 449-456. [34] D. Gunther, R. Frischknecht, C.A. Heinrich, H.J. Kahlert. Capabilities of an argon fluoride 193 nm excimer laser for laser ablation inductively coupled plasma mass spectometry microanalysis of geological materials, J. Anal. At. Spectrom. 12 (1997) 939-944.

PAGE 156

156 [35] H. Lindner, J. Koch, K. Niemax. Production of ultrafine particles by nanosecond laser sampling using orthogonal prepulse laser breakdown, Anal. Chem. 77 (2005) 7528-7533. [36] R.S. Houk, N. Praphairaksit. Dissociation of polyatomic ions in the inductively coupled plasma, Spectrochim. Acta Part B 56 (2001) 1069-1096. [37] D.A. Wilson, G.H. Vickers, G.M. Hieftje. Ionization temperatures in the inductively coupled plasma determined by mass-spectrometry, Appl. Spectrosc. 41 (1987) 875-880. [38] S.A. Lehn, K.A. Warner, M. Huang, G.M. Hieftje. Effect of an inductively coupled plasma mass spectrometry sampler interface on electron temperature, electron number density, gas-kinetic temperature and analyte emission intensity upstream in the plasma, Spectrochim. Acta Part B 57 (2002) 1739-1751. [39] B. Budic. Matrix effects in inductively coupled plasma atomic spectrometry using an ultrasonic nebulizer, J. Anal. At. Spectrom. 13 (1998) 869-874. [40] J.W. Olesik, Investigating the fate of individual sample droplets in inductively coupled plasmas, App. Spectrsc. 1 (1997). [41] K. Warner, G.M. Hieftje, Thomson scattering from analytical plasmas, Spectrochim. Acta Part B 57 (2002) 201-241 [42] K. H. Becker, U. Kogelschatz, K. H. Schoenbach, R. J. Barker, NonEquilibrium Air Plasmas at Atmospheric Pressure, CRC press (2004) [43] H. R. Griem, Spectral Line Broadening by Plasmas, Academic Press, New York (1974). [44] A. Ashkin, J.M. Dziedzic, J.E. Bjorkholm and S. Chu, Observation of a SingleBeam Gradient Force Optical Trap for Dielectric Particles.,Opt. Lett. 11 (5) (1986) 288-290. [45] Ashkin, A. "History of Optical Trapping and Manipulation of Small-Neutral Particle, Atoms, and Molecules." IEEE Journal of Selected Topics in Quantum Electronics 6(6) (2000) 841-856. [46] Lang, M.J., Fordyce, P.M. and Block, S.M. Combined optical tweezers and single-molecule fluorescence. J. Biology 2:6 (2003) [47] ES Simpson, GA Lithgow, SG Buckley, Three-dimensional distribution of signal from single monodisperse aerosol particles in a laser induced plasma: initial measurements, Spectrochim. Acta Part B 62 (2007) 1460-1465.

PAGE 157

157 [48] J.E. Carranza, K. Iida and D.W. Hahn, Conditional data processing for singleshot spectral analysis by use of laser-induced breakdown spectroscopy, Applied Optics 42 (2003) 6022-6028. [49] J.E. Carranza and D.W. Hahn, Plasma volume considerations for analysis of gaseous and aerosol samples using laser induced breakdown spectroscopy, J. Analytical Atomic Spectr., 17 (2002) 1534-1539. [50] D.W. Hahn, W.L. Flower, and K.R. Hencken. Discrete particle detection and metal emissions monitoring using laser-induced breakdown spectroscopy Applied Spectroscopy 51 (1997) 1836-1844. [51] The Dust Hypothesis for Caribbean Coral Bleaching, United States Geological Survey. [52] K. Iida C.Y. Wu, and D.W. Hahn, In situ analysis of the interaction of manganese vapor and silica microspheres in a laboratory-scale combustor, Combust. Sci. and Tech., 176 (2004) 453-480. [53] V. Hohreiter and D.W. Hahn. Calibration effects for laser-induced breakdown spectroscopy of gaseous sample streams: Analyte response of gaseous phase species vs. solid phase species, Anal. Chem. 77 (2005) 1118-1124. [54] V. Hohreiter, A. Ball, D.W. Hahn. Effects of aerosols and laser cavity seeding on spectral and temporal stability of laser-induced plasmas: application to LIBS, J. Anal. At. Spectrom. 19 (2004) 1289-1294. [55] D.W. Hahn, J.E. Carranza, G.R. Arsenault, H.A. Johnsen, and K.R. Hencken. Aerosol generation system for development and calibration of laser-induced breakdown spectroscopy instrumentation, Review of Scientific Instruments, 72 (2001) 3706-3713. [56] D.W. Hahn and M.M. Lunden. Detection and analysis of aerosol particles by laser-induced breakdown spectroscopy, Aerosol Science and Technology, 33 (2000) 30-48. [57] P.K. Diwakar, P.B. Jackson, D.W. Hahn. Investigation of Multi-component Aerosol Particles and the Effect on Quantitative Laser-induced Breakdown Spectroscopy: Consideration of Localized Matrix Effects, Spectrochimica Acta Part B, 62 (2007) 1466-1474 (2007). [58] D.W. Hahn, W.L. Flower, and K.R. Hencken. Discrete particle detection and metal emissions monitoring using laser-induced breakdown spectroscopy, App. Spec. 51 (1997) 1836-1844. [59] M. Essien, L.J. Radziemski, and J. Sneddon. Detection of cadmium, lead and zinc by laser-induced breakdown spectroscopy, J. Anal. Atomic Spectrom. 3 (1988) 985-988.

PAGE 158

158 [60] B.C. Windom, P.K. Diwakar, D.W. Hahn, Dual-Pulse LIBS for Analysis of Gaseous and Aerosol Systems: Plasma-Analyte Interactions, Spectrochim. Acta Part B 61 (2002) 788-796. [61] D. Mukherjee, A. Rai and M.R. Zachariah, Quantitative laser-induced breakdown spectroscopy for aerosols via internal calibration: application to the oxidative coatings of aluminum nanoparticles, Aerosol Sci. 37 (2006), 677695. [62] E. Tognoni, M. Hidalgo, A. Canals, G. Cristoforetti, S. Legnaioli, A. Salvetti, V. Palleschi. Robust or non-robust conditions in Inductively Coupled Plasma Atomic Emission Spectroscopy measurements: Is it possible to improve the reliability of the monitoring, Spectrochim. Acta Part B, 78 (2006) 1509-1514. [63] M. Miclea, I. Exius, H. Lindner, C. Garcia, and K. Niemax. Characterization of laser produced particles by ICP-OES, presented at the 8th European Workshop on Laser Ablation, Zurich, July 19-21, 2006. [64] J. Aubreton, M. F. Elchinger, P. Fauchias. New method to calculate thermodynamic and transport properties of a multi-temperature plasma: application to N2 plasma, Plasma Chem. And Plasma Processing 18 (1998) 1-27. [65] P. S. Dalyander, I.B. Gornushkin, D.W. Hahn, Numerical Simulation of LaserInduced Breakdown Spectroscopy: Modeling of Aerosol Analysis with Finite Diffusion and Vaporization Effects, Spectrochimica Acta Part B, 63 (2008) 29330. [66] E. E. Salpeter, Electron densities fluctuations in a plasma, Phys. Rev., 120 (1960) 1528-1535 [67] D.E. Evans, J. Katzenstein, Laser light scattering in laboratory plasmas, Rep. Prog. Phys., 32, (1969) 207-271. [68] M. Huang, GM. Hieftje, Thomson scattering from an ICP, Spectrochim. Acta Part B, 40 (1985) 1387 [69] M.D.Bowden, M.Kogano, Y.Suetome, T.Hori, K Uchino, K.Muraoka, Comparison of electron property measurements in an inductively coupled plasma made by Langmuir probe and laser Thomson scattering techniques, J.Vac. Sci. Technol. A 17 (1999) 493. [70] D.M. Villeneuve, H. A. Baldis, J.E. Bernard, R. Benesch, Collective Thomson scattering in a laser-produced plasma resolved in time, space, frequency, or wave number, J. Opt. Soc. Am. B 8 (1991) 895-902. [71] J.E. Bernard, H.A. Baldis, D.M. Villleneuve, K. Estabrook, Time resolved Thomson scattering measurements of the electron and ion temperatures in a high intensity laser-plasma interaction, Phys. Fluids, 30 (1987) 3616-3623.

PAGE 159

159 [72] P.K. Diwakar, D.W. Hahn, Study of Early Laser-induced Plasma Dynamics: Transient Electron Density Gradients via Thomson Scattering and Stark Broadening, and the Implications on LIBS Measurements, Spectrochimica Acta Part B, 63 (2008) 1038-1046. [73] NIST database, http://physics.nist.gov/PhysRefData/ASD/index.html [74] P.D. Marker, R.W. Terhune, C.M. Savage, Proc. 3rd Int. Conf. on Quantum Electronics ,Grivet and N Bloembergen, (1963) 1559-1565. [75] C.G. Morgan, Laser-induced breakdown of gases, Rep. Prog. Phys., 38 (1975) 621-665. [76] J.L. Beduneau, Y. Ikeda, Spatial characterization of laser-induced sparks in air, J. Quant. Spec. Rad. Transfer, 84 (200) 123-139. [77] Y.-L. Chen, J.L. Lewis, C. Parigger, Spatial and temporal profiles of pulsed laser-induced air plasma emissions, J. Quant. Spec. Rad. Transfer, 67 (2000) 91103. [78] F.V. Bunkin, M.V. Fedrov, Bremsstrahlung in a strong radiation field, Sov. Phys. JETP, 22 (1966) 844-847. [79] C.M. Ireland, C.G. Morgan ,Gas breakdown by a short laser pulse, J. Phys. D: Appl. Phys 6 (1973) 720-729 [80] J.M. Aaron, C.M. Ireland, G. Morgan, Aberration effects in the interaction of focused laser beams with the matter, Phys D: Appl. Phys.,7 (1974) 1907. [81] M.H. Key, D. A. Preston, T.P. Donaldson, Self focusing in gas breakdown by laser pulses, J. Phys. B: At. Mol. Phys. 3: L88 (1970). [82] J.W. Olesik, 2003 meeting of the federation of analytical chemistry and spectroscopy socities, FACCS, Fort Lauderdale, FL, October 2003. [83] P.C. Chylek, M.A. Jarzembski, V. Srivastava, R.G. Pinnick, J.D Pendleton, J.P. Cruncleton, Effect of spherical particles on laser-induced breakdown of gases, App. Optics, 26 (1987) 760-762. [84] G.M. Hieftje, H.V. Malmstadt, A unique system for studying flame spectrometric processes, Anal. Chem.,40 (1968) 1860-1867. [85] H.A. Archontaki, S.R. Crouch, Evaluation of an isolated droplet sample introduction system of laser-induced breakdown spectroscopy, App. Spec., 42 (1988) 741-746.

PAGE 160

160 [86] S. Groh, C.C. Garcia, A. Murtazin, V.Horvatic, K. Niemax, Local effects of ataomizing analyte droplets on the plasma parameters of the inductively coupled plasma, Spec. Acta B, 64 (2009) 247-254. [87] C. Janzen, R. Fleige, R. Noll, H. Schwenke, W. Lahmann, J. Knoth, E. Jantzen, A. Oest, P. Koke, Analysis of small droplets with a new detector for liquid chromatography based on laser-induced breakdown spectroscopy, Spec. Acta B, 60 (2005) 993-1001. [88] M. Englmann, A. Fekette, I. Gebefugi, P. S. Kopplin, The dosage of small volumes for chromatographic quantifications using a drop-on-demand dispenser system, Anal Bioanal Chem., 388 (2007) 1109-1116 [89] P. Jacob, A. Stockhaus, R. Hergenroder, D. Klockow, Phase transfer and freezing processes investigated on acoustically levitated aqueous droplets, Fres J. Anal. Chem. 371 (2001) 726-733 [90] K. Lo, N. Cheung, ArF laser-induced plasma spectroscopy for part-per-billion analysis of metal ions in aqueous solutions, App. Spec., 56 (2002) 682-688. [91] R. Knopp, F.J. Scherbaum, J.I.Kim, Laser-induced breakdown spectroscopy (LIBS) as an analytical tool for the detection of metal ions in aqueous solutions, Fres. J. Anal Chem. 355 (1996) 16-20. [92] D. A Cremers, L.J. Radziemski, T.R. Loree, Spectrochemical analysis of liquids using the laser spark, Appl. Spec., 38 (1984) 721-729. [93] S.R. Turns, An Introduction to Combustion: Concepts and Applications (2nd Ed.), MacGraw Hill, Boston (2000). [94] R.A.Svehla,Estimated viscosities and thermal conductivities of gases at high temperatures, NASA Tech. Rep. R-132 (1962) [95] R.Giro,, M.J. Caldas, Calcium deposition on poly(para-phenylene vinylene): Molecular dynamics simulations, Phys. Rev. B, 76 (2007) 161303. [96] K.B. Tarmyshov, F.M.Plathe. Ion bonding to cucurbit[6] uril: structure and dynamics,J.Phys. Chem. B, 110 (2006) 14463-14468. [97] N. Yoshida, S. Phongphanphanee, F. Hirata, Selective ion bonding by protein probed with the statistical mechanical integral equation theory, J. Phys. B. 111 (2007) 4588-4595.

PAGE 161

161 BIOGRAPHICAL SKETCH Prasoon Diwakar was born on 21st of June, 1979 in Allahabad, India to Bhupati Kishore and Sushila Devi. Prasoon grew up in the Satna and Shahdol districts in central part of India. He attended high school at Good Shepherd Convent School in Shahdol district. After graduation from high school, he competed for joint entrance examinations for the Indian Institute of Technology and got selected for bachelors program in mechanical engineering at Indian Institute of Technology, Kanpur, India. During his undergraduate program, he worked at TVS Suzuki motors, India and CNR-TEMPE, Milano as summer intern which motivated him to pursue research work in the future. He received his bachelor of technology degree from IIT Kanpur in 2003. Following graduation, he got admitted to PhD. program in mechanical engineering at the University of Florida in 2004. He earned his master of science degree (non-thesis) from the University of Florida in August 2006.The work presented here is a culmination of the research carried out during his PhD. studies at the University of Florida.