<%BANNER%>

Implementation of Aerodynamic Focusing and Dual-Pulse Configuration to Improve Laser-Induced Breakdown Spectroscopy Aero...

HIDE
 Title Page
 Acknowledgement
 Table of Contents
 List of Tables
 List of Figures
 Abstract
 Introduction
 Experimental methods
 Results and discussion
 Conclusions and proposed future...
 Appendices
 References
 Biographical sketch
 

PAGE 1

IMPLEMENTATION OF AERODYNAM IC FOCUSING AND A DUAL-PULSE CONFIGURATION TO IMPROVE LASER-INDUCED BREAKDOWN SPECTROSCOPY AEROSOL PARTICLE SAMPLING RATES AND ANALYTE RESPONSE By BRET C. WINDOM A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 By Bret C. Windom

PAGE 3

ACKNOWLEDGMENTS First and foremost, I would like to thank Dr. Hahn for his guidance and leadership provided during the current study. Because of hi s enthusiasm and expertise in the subject I have learned alot in the short time I have been here. Secondly, I would like to thank my lab mates for providing assistance in my resear ch and course work, and also, for creating an enjoyable and stimulating work atmosphere. iii

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES.............................................................................................................vi LIST OF FIGURE S..........................................................................................................vii ABSTRACT.......................................................................................................................xi CHAPTER 1 INTRODUCTION...........................................................................................................1 1.1 Overview.................................................................................................................1 1.2 Laser Induced Breakdown Spectroscopy (LIBS) Process......................................2 1.2.1 Plasma Formation.........................................................................................2 1.2.2 Temporal Evolution and Decay of a Laser-Induced Plasma........................3 1.2.3 Atomic Spectral Em ission Collection..........................................................5 1.3 LIBS Signal............................................................................................................5 1.3.1 Analyte Response.........................................................................................5 1.3.2 Calibration....................................................................................................6 1.3.3 Calibration Curve.........................................................................................7 1.3.4 Limits of Detection.......................................................................................9 1.4 Aerosol Based LIBS.............................................................................................10 1.4.1 Mass Concentration and Size Detection.....................................................10 1.4.2 Continuous Emission LIBS........................................................................11 1.4.3 Conditional Analysis..................................................................................12 1.4.4 Fundamental Aerosol Studies.....................................................................13 1.5 Dual-Pulse LIBS...................................................................................................15 1.6 Objective...............................................................................................................17 2 EXPERIMENTAL METHODS.....................................................................................19 2.1 LIBS Experimental Setup and Procedure.............................................................19 2.1.1 Plasma Creation Methods...........................................................................19 2.1.2 Emission Collection....................................................................................20 2.1.3 Aerosol Generation.....................................................................................21 2.2 Dual-Pulse LIBS: Gas and Aerosol Study............................................................24 2.2.1 Laser Configuration....................................................................................24 iv

PAGE 5

2.2.2 Experimental Analyte Generation..............................................................26 2.2.3 Spectral Measurements...............................................................................27 2.2.4 Transmission Measurements......................................................................29 2.3 Particle Lens Study...............................................................................................29 2.3.1 Laser Configuration....................................................................................29 2.4.2 Analyte Generation.....................................................................................30 2.4.3 Spectral Measurements...............................................................................32 2.4.4 Particle Counting Measurements................................................................32 3 RESULTS AND DISCUSSION....................................................................................33 3.1 Dual-Pulse LIBS Study.........................................................................................33 3.1.1 Transmission Experiments.........................................................................33 3.1.2 Spectral Analysis of Gaseous Analyte.......................................................38 3.1.3 Spectral Analysis of Fine Ca lcium-Based Aerosol Analyte.......................47 3.1.4 Spectral Analysis of Boro silicate Glass Microspheres...............................53 3.2 Particle Lens Study...............................................................................................56 3.2.1 First Attempts.............................................................................................56 3.2.2 Sheath Flow Experimental Results.............................................................58 3.2.3 Summary and Particle Counting Results....................................................62 4 CONCLUSIONS AND PR OPOSED FUTURE WORK...............................................66 APPENDIX A COMPONENTS OF SPE CTROSCOPIC SYSTEMS..................................................68 B PARTICLE LENS CALCULATIONS.........................................................................70 LIST OF REFERENCES...................................................................................................71 BIOGRAPHICAL SKETCH.............................................................................................75 v

PAGE 6

LIST OF TABLES Table page 3-1 Data representing the analyte response (Ca II) and the hit rates for the two aerosol injection methods.........................................................................................56 3-2 Spectra and hit data comparing the tube to the particle lens..............................59 A-1 List of components used in the dual-pulse study.....................................................68 A-2 List of components used in the particle lens study...................................................69 B-1 Results of calculations used to determine predicted sampling rates for the particle lens and steel tube..................................................................................70 vi

PAGE 7

LIST OF FIGURES Figure page 1-1 Spectral line representation with increasing atomic concentration (a h). Once the concentration reaches a certain amount the emission intensity reaches a limit given by Bb m (blackbody radiator). The wings then start to widen causing calibration curves to flatten [Ingle and Crouch 1988]................................................8 1-2 Calibration curve of growth. At first, th e curve increases linearly until a certain concentration is achieved and the maximu m linear atomic emission intensity is reached, equal to that of a blackbody radiator. At this instance self-absorption begins, and the curve begins flatten out.....................................................................9 1-3 Illustrations representing two setups used in many of the dual-pulse studies; [A] An orthogonal configuration and [B ] An in-line configuration...............................16 2-1 Optical setup for the delivery of the laser irradiance to the sample volume resulting in the creation of a plasma.........................................................................19 2-2 Optical setup for the collection of plasma emission from the sample volume.........21 2-3. General schematic of the aerosol generation system. The co-flow rates were those used in the dual-pulse study [Hahn 2001]......................................................22 2-4 Chart displaying Cu analyte counts of re mnant solutions from a nebulizer as a function of nebulization time. Error bars represent the standard deviation.............23 2-5 Experimental apparatus for single an d dual-pulse LIBS configurations..................24 2-6 Schematic representing the trigger se tup to achieve laser pulses fired simultaneously in time or at a variable delay. Delay generator 1 set to 77.75 s would result in the lasers firing simultane ously. Adjusting the variable delay of generator 1 would result in delaying laser 2 pulse...................................................26 2-7 The optical setup (top view) for the partic le lens experiments. Like the dualpulse setup, the plasma emission was b ack collected. The aerosols were drawn through the particle lens into the plasma by a vacuum pump situated on the opposite side of the 5-way cross..............................................................................30 vii

PAGE 8

2-8 Simple schematic of the particle lens. Ae rosols flow through a series of radial decreasing concentric tubes before exiti ng through an outlet causing a streamline exit into the laser induced plasma............................................................................31 3-1 Laser 2 transmission as a function of dualpulse laser delay times for the pure air and fine calcium aerosol sample stream s. The horizontal line represents the transmission of Laser 2 alone (i.e., single pul se LIBS). Note the plot is linear for delay times less than 100 ns.....................................................................................34 3-2 Spectra showing the two nitrogen atom ic emission lines at 491.4 nm and 493.5 nm. The spectra correspond to dual-pul se LIBS with 500 ns delay (lower spectrum) and single-pulse (Laser 2 only) LIBS (upper spectrum). Both spectra have the same scale..................................................................................................38 3-3 The 493.5-nm nitrogen emission line peak-to-base ratio measurements for the pure air sample as a function of dual-pulse laser delay times. The P/B ratios are also shown corresponding to the Lase r 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizontal line re presents the average P/B ratio for Laser 2 only. A representative error bar is in cluded on the 100 ns Laser 1 only mark.....40 3-4 The 493.5-nm nitrogen emission line signal-to -noise ratio measurements for the pure air sample as a function of dual-pulse laser delay times. The SNR are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizontal line represents the average SNR ratio for Laser 2 only.......41 3-5 Spectra showing the oxygen (I) triplet at 394.7 nm. The spectra corresponds to dual-pulse LIBS with 500 ns delay and si ngle-pulse (Laser 2 only) LIBS. Both spectra have the same scale......................................................................................42 3-6 Oxygen peak-to-base measurements for th e filtered air sample as a function of dual-pulse laser delay times. The average peak-to-base ratio for Laser 2 only is represented by the dashed horizontal line................................................................43 3-7 The 397.4-nm oxygen emission line signal-to-noise ratio measurements for the pure air sample as a function of dual-pulse laser delay times. The S/N are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizontal line represents th e average S/N ratio for Laser 2 only.........44 3-8 Image detailing the proposed mechanism, in which an expanding plasma volume will force the nano-scale gas phase particles outward depleting their concentration in the plasma core [Hohreiter and Hahn 2005a]................................46 3-9 Spectra showing the Ca II atomic em ission lines at 393.4 and 396.9 nm for both the dual-pulse configuration with a 250-ns delay, and for Laser 2 only. Both spectra have the same intensity scale, and the dual-pulse spectrum has been shifted upward by 400 counts for clarity..................................................................47 viii

PAGE 9

3-10 Spectra showing the Ca II atomic em ission lines at 393.4 and 396.9 nm for both the dual-pulse configuration with a 750-ns delay, and for Laser 2 only. Both spectra have the same intensity scale, and the dual-pulse spectrum has been shifted upward by 400 counts for clarity..................................................................48 3-11 Spectra showing the Ca II atomic em ission lines at 393.4 and 396.9 nm for both the dual-pulse configuration with a 50 -s delay, and for Laser 2 only. Both spectra have the same intensity scale.......................................................................49 3-12 The 393.4 -nm calcium II emission line peak -to-base ratio measurements for the fine calcium aerosol sample as a functi on of dual-pulse laser delay times. The P/B are also shown corresponding to the La ser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizontal line represents the average P/B ratio for Laser 2 only. 50 3-13 The 393.4 -nm calcium II emission line sign al-to-noise ratio measurements for the fine calcium aerosol sample as a f unction of dual-pulse laser delay times. The SNR are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizontal line represents the average SNR for Laser 2 only..............................................................................................................51 3-14 This image depicts the basis for the in crease analyte response in a dual-pulse configuration for an aerosol sample. The la rger aerosol particles resist the radial force exerted on them by the expandi ng plasma volume created by laser 1 and remain inside the core while the smaller gas phase particle are expelled, thereby creating a more aerosol con centrated volume that is awaiting the plasma from laser 2 [Hohreiter and Hahn 2005a].........................................................................52 3-15 Images depicting the calcium aerosol lig ht scattering of a green diode laser. Looking down into the sample chamber, th e beam path runs directly across the center of the particle beam which is ali gned in the center of the annular tube. The green line represents the presence of particles. The two images show the difference in particle location without [l eft] and with [right] the annular sheath flow. As you can see, marked by the small s pot of scattered light at the center of the tubes, the image with the annular sh eath flow [right] ef fectively separated the particle exiting the lens with those that were re-c irculating in the chamber......58 3-16 Histogram categorizing the peak-to-base ra tio of each individual hit for the tube and the particle lens. The overall inte nsity is down for the particle lens due to the smaller hit rates, but the distribution for both methods follow very similar trends. 60 3-17 Probability plot comparing the distribution of the P/B ratios for all the individual hits collected after the conditional an alysis approaches were performed................61 3-18 The thin walled tube [A] allowed fo r the maximum effective plasma volume, represented by the shaded region, to ex cite the aerosol samples, while the ix

PAGE 10

particle lens [B] reduced the effectiv e plasma volume which aided in reducing the particle hit rates..................................................................................................64 x

PAGE 11

Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science IMPLEMENTATION OF AERODYNAM IC FOCUSING AND A DUAL-PULSE CONFIGURATION TO IMPROVE LASER-INDUCED BREAKDOWN SPECTROSCOPY AEROSOL PARTICLE SAMPLING RATES AND ANALYTE RESPONSE By Bret C. Windom December 2006 Chair: David Hahn Department: Mechanical Engineering This study focused on two alternative methods to increase the analyte response using laser induced breakdown spectroscopy (L IBS) specifically on gaseous and aerosol phase analytes. The first, using a dual-pulse LIBS configuration to enhance analyte response of elements in an air sample and aerosol particles consisting of calcium; and second, using a particle focusing lens as a way to feed aerosols into the laser-induced plasma to produce higher particle hit rate s and/or an enhanced analyte response. Dual-pulse LIBS has previously demonstr ated to significantly enhance the analyte peak-to-base and signal-to-noise ratios for solid and liqui d phase analytes. This study focused on the effects of an orthogonal dual-pulse laser configuration on the atomic emission response for both purely gaseous a nd calcium-based aerosol samples. The gaseous sample consisted of purified (i.e., aerosol free) air, from which nitrogen and oxygen spectral emission lines were analyzed. Measurements for the gaseous system xi

PAGE 12

resulted in no notable improvements with th e dual-pulse configuration as compared to single-pulse LIBS. Experiments were also conducted in purifie d air seeded with calciumrich particles, which revealed a marked im provement in calcium atomic emission peakto-base (~2-fold increase) and signal-to-nois e ratios (~4-fold increase) with the dualpulse configuration. In addition to increased analyte response, dual-pulse LIBS yielded an enhanced single-particle sampling rate when compared to conventional LIBS. Transmission measurements with respect to the plasma-creating laser pulse were recorded for both single and dual-pulse methods over a range of temporal delays. In consideration of the spectroscopic and transm ission data, the plasma-a nalyte interactions realized with a dual-pulse methodology are explai ned in terms of the interaction with the initially expanding plasma, which differs betw een gaseous and particulate phase analytes. A particle-focusing lens is a way to redu ce the cross-section of the flow and eject a narrow streamline of particles. This study fo cused on the notion that if a streamline of particles could be injected directly into the center of a laser-induced plasma by a particle lens, higher aerosol sampli ng rates and better analyte re sponse could be achieved. Following all experiments and analysis, it was de termined that a particle lens with exit cross-section diameter of 0.0295 showed a decrease in particle hit rates and an unchanged analyte response when compared to th at of a diameter standard thin walled tube. The average hit rates decreased about 75% with the particle lens and resulting analyte response of all individua l hits for the two injection methods were nearly the same. The lower particle hit rate with the focusing le ns was partially attributed to a reduction in particle transmission (as measur ed with a particle counter), while other effects such as changes in the effective plasma sampling volume were also considered. xii

PAGE 13

CHAPTER 1 INTRODUCTION 1.1 Overview Laser-induced breakdown spectroscopy (LIB S) also referred to as laser-induced plasma spectroscopy (LIPS) had its inception in the 1960s with the development of the first laser. As instrumentation progressed and became less expensive, LIBS grew in popularity as an analytical technique. Since the 1980s research and published literature has increased, including many literature reviews [Schechter 1997, Smith 2001, Tognoni 2002, Radziemski 2002, Lee 2004, Winefordner 2004], as LIBS has begun to obtain recognition on the likes of other analytical techniques such as inductively coupled plasma-mass spectrometry (ICP-MS) and inductively coupled plasma-atomic emission spectrometry (ICP-AES) [Winefordner 2004]. A LIBS system entails tightly focusing a pulsed laser beam onto the medium of interest, in which the resulting increase in ir radiance acts in disasso ciating molecules and creating a volume with high ion and free electron densities along with high temperatures (in excess of 30,000 K) called a plasma or spark. As the plasma temperatures cool and the ionization degree falls, the electronically ex cited atoms within the plasma relax and emit elemental specific radiation (i.e., atomic emission) which can be analyzed with a spectrometer in combination with an intens ified charge coupled device (ICCD) detector. The measured spectral lines can be used to carry out measurements relating species existence, concentration, and mass. The LI BS technique has the ability to perform 1

PAGE 14

2 analysis on gases, liquids, solids, and, an im portant research topic, aerosol systems. Because of simple and relatively inexpensiv e instrumentation, its aptness for real-time in situ measurements, and its prospect to being a pplied to micro-systems, LIBS shows huge promise to becoming a prominent analytical technique in the future. 1.2 Laser Induced Breakdown Spectroscopy (LIBS) Process 1.2.1 Plasma Formation A laser-induced plasma, qualitatively ch aracterized by a bright spark followed by an audible shock wave, is created when th e irradiance of the laser beam exceeds the dielectric strength or the br eakdown threshold of the medium [Weyl 1989]. For example, air has a breakdown threshold that can range from 90-1600 GW/cm2 for a 1064 nm laser source, and from 30-380 GW/cm2 for 266 nm lasers [Smith 2001]. Beam irradiance can be increased by focusing the beam to its diffraction limit, by using short duration pulses (~ ns to fs), and by simply increasing the laser pulse energy. The two mechanisms that primarily result in plasma formation are multi-photon ionization and cascade ionization, also known as avalanche ionization. Multi photon ionization is the process when electrons are excited by some incident radiation source, to the point that their energy becomes great enough to break away from the atom and become free electr ons. Typically, for the pr ocess to occur with visible and IR radiation souces, the energy from multiple photons are needed, therefore the probability for occurrence is based on photon density. Multi photon ionization can be described in the following equation. M + n(h ) e+ M+ (1.1) In which M is the neutral atom and n(h ) is n number of photons when combined create a free electron and an ionized atom. The numbe r of photons needed, n, is dependent on the

PAGE 15

3 energy of a single photon. For example, since a photon with a wavelength of 532 nm is more energetic than one of 1064 nm (2.33 eV vs. 1.17 eV), the 532 nm light would be able to achieve ionization with less photons than that of 1064 nm light (i.e., in order to ionize N2, n532 = 4 photons, while n1064 = 8 photons). The second mechanism responsible for plasma creation is cascade ionization. Since free electrons are not quantized, the radiative energy absorbed from the incident laser by the free electrons are converted into translational energy which causes them to collide with neutral atoms or molecules re sulting in their ioni zation and another free electron, represented in the following equation. e+ M 2e+ M+ (1.2) The relationship above shows that the free electrons are doubled with every occurrence of cascade ionization, therefore causing an exponentia l increase in the number of free electrons. Cascade ioni zation is thought to be the prominent mechanism in plasma creation, but since free electrons are needed, mu lti photon ionization is sa id to initiate the breakdown process. However, due to th e higher energy per photon, for shorter wavelength laser sources (<1 m), such as the 266 nm (4.67 eV)source, the process of multi photon ionization is more important in th e plasma creating process, whereas it takes a back seat to cascade ionization when deali ng with longer waveleng th sources (i.e., 1064 nm) [Martin and Cheng 2000]. 1.2.2 Temporal Evolution and Decay of a Laser-Induced Plasma Although plasma characteristics can vary based on laser power, laser wavelength, optical setup, sample material and concentr ation, the following represents the evolution and life of a typical laser-induced plasma By the end of the laser pulse, plasma temperatures and free electron number densities rise to their maximum, as much as

PAGE 16

4 40,000 K and 101819/cm3, respectively, as approximately 50% to 90% of the laser pulse energy is coupled into the plasma. It should be noted that radiant emission from the plasma (also known as the continuum radiation) in the ultraviolet-visible range occurs shortly after the laser pulse reaches the focal s pot and is representati ve of emission from a blackbody. The spectral distribution of the contin uum can vary with plasma temperature, for example, a hotter plasma will shift the continuum to smaller wavelengths and vice versa for cooler plasmas. Continuum emission is the result of two processes, recombination (free-bound) and Bremsstrahlung (free-free). Recombination occurs as an ion captures a free-electron resulting in emission of the excess energy as the fr ee electron becomes bound. Bremsstrahlung emission occurs as a free electr on decelerates near and then past an ion causing an energy adjustment resulting in the emission of a photon equal to the energy change. Generally, plasma temperatures and free el ectron densities peak by the tail end of the laser pulse, although the plasma continue s to grow in volumet ric size and continues its strong broadband continuum emission, which dominates atomic lines early in time. After a few microseconds the plasma has signifi cantly cooled due to energy transfer from radiation, recombination, and quenching, a nd the continuum effects become less dominate, therefore spectral atomic lines b ecome detectable. Since the bound electrons are quantized, as they start to relax to electronic ground st ates, photons, with frequencies inherent to each atom present in the plasma volume are emitted (similar to laser induced fluorescence). The optimal delay after the lase r pulse in which atomic lines are best

PAGE 17

5 resolved over the continuum emission is elem ental dependent and can range from 1 to 30 s (Buckley 2000). After 30 to 50 s, the neutral atoms start to combine, creating molecules which can emit at their own emission bands. Emissi on can occur up to about one hundred microseconds after the initial pul se, but is hard to detect af ter about 50 s due to the low signal levels. After a couple hundred microseconds, the plasma has essentially dissipated and returned to near ambient conditions. 1.2.3 Atomic Spectral Emission Collection Since light from the plasma emits equally in all directions, th e collection devices can be arranged through multiple configur ations, but certain ones do have their advantages. In most cases the emitted light is focused onto the tip of a fiber optic bundle and transmitted to a spectrometer in which th e light is diffracted a nd dispersed so that a specific wavelength range of the emitted light is analyzed. The dispersed light is then applied to the surface of an intensified charged coupled device (ICCD) creating a spectrum, so that intensities of specific wa velengths representing species in the plasma can be analyzed. 1.3 LIBS Signal 1.3.1 Analyte Response When analyzing spectra, one can qualitativ ely determine the existence of certain elements through the existence of atomic spectra l lines, but this is onl y a fraction of what makes LIBS a robust technique. It is known that the intensity of the atomic signal is proportional to the number of emitting atoms/ions. This is important because it allows LIBS to be a quantitative technique to measure analyte concentration.

PAGE 18

6 So what is the best way to represent signal intensities so that concentration measurements can be extracted? Rather than measuring the absolute intensity, which is the combination of the background (continuum em ission) and the atomic signal, a better and more widely used measurement include s taking the integrat ed peak above the continuum and normalizing it by the intensity of the continuum [Hahn 1997]. This ratio is called the peak-to-base (P/B) a nd is advantageous due to the ability to minimize shot-toshot variations by normaliza tion to the continuum [Carranza and Hahn 2002a]. Another popular measurement among LIBS researchers to quantify analyte response is the signalto-noise ratio (S/N). In this measurement, the analyte signal, which may be the full width integrated emission peak area, is divided by the spectral noi se, which may be represented by the root-mean-square (RMS) of the continuum intensity adjacent to each peak. The RMS describes the average square of the devi ation of the continuum from a linear fit of itself, and is representative of the smoothness of the continuum region, which therefore is a measure of analyte signal limitation [Ingle and Crouch 1988]. 1.3.2 Calibration In most cases when LIBS is applied to detecting amounts and concentrations of certain species within a medium, calibration mu st be made, with th e exception attributed to studies such as those by Ciucci [1999] and Bilajic [2002] in which calibration free LIBS procedures were developed. Calibrati on methods include analyzing spectra from standards, which contain varying concentrations of analyte within a medium identical as possible to that of the sample for which the concentration information is sought. By using the same material in the standards as in the sample of interest (i.e., matching the matrix), plasma matrix effects can be avoided. In gene ral, matrix effects are alterations in analyte emission due to the changes in plasma char acteristics such as temperature, electron

PAGE 19

7 density, quenching of atomic spectral lines by other species, and the loss of emitting elements due to recombination. For example, significance of these effects was shown in a study done by Gleason and Hahn [2001], in whic h mercury spectral lines were shown to differ as the concentration of oxygen in the surrounding gas was altered. They concluded that the reduction in the reco rded mercury emission was due to molecular oxygen species, mainly O2 and NO, quenching the mercury atomic signal. Also, Ismail [2004] demonstrated such plasma matrix effects, s howing that the detection limits of manganese, silicon, magnesium, and copper varied when they were embedded in an aluminum standard versus a steel standard. 1.3.3 Calibration Curve Calibration curves that are linear are preferred when performing concentration analysis using LIBS. However, nonlinea rities are not uncommon, and do occur for various reasons such as exceeding the concen tration limit, quenching, and recombination. Mentioned before, as the concentration of th e analyte increases so too does the intensity of the atomic signal. As the emission inte nsity of the analyte grows, it can reach an intrinsic limit equivalent to the emission of a blackbody radiator from Planks law. This then causes the wings of the atomic line to widen, Figure 1-1, and can be thought in terms of line broadening or self-absorption. At low analyte concentrations, the signal response grows proportionally with the number of analyte atoms. When the atomic concentration becomes too high, the emission irradiance of a blackbody emitter is reached, a nd subsequently an increasing fraction of the emitted radiation is self-absorbed by atoms in lower energy states, and the curve starts

PAGE 20

8 Figure 1-1. Spectral line re presentation with increasi ng atomic concentration (a h). Once the concentration reaches a certa in amount, the emission intensity reaches a limit given by Bb m (blackbody radiator). The wings then start to widen causing calibration curves to flatten [Ingle and Crouch 1988]. to level out, as seen in Figure 1-2 [Ingle a nd Crouch 1988]. Therefore, it is important to perform LIBS measurements in the concentration range th at gives a linear response, which is typically on the order of under 10%, indicated by the concentration region to the left of the dotted line in Figure 1-2. Other sources that affect the calibration curve of growth include quenching and recombination. Quenching is a non-radiative decay of atomic emission. It is species specific and is a function of the analyte concen tration as well as the concentrations of the constituents of the plasma matrix. Recombin ation, like quenching, is species specific and is also a function of analyte concentration. It occurs when the analyte and another matrix containing species have a high affinity for each other resulting in non-radiative energy

PAGE 21

9 Concentration Signal Response (P/B or S/N) Saturation Regime Figure 1-2. Calibration curve of growth. At first, the curve increases linearly until a certain concentration is achieved an d the maximum linear atomic emission intensity is reached, equal to that of a blackbody radiator. At this instance self-absorption begins, and th e curve begins flatten out. loss due to their combination (e.g., Hg + O HgO). The last two sources of nonlinearity may or may not be important, but should be at least considered when investigating matrix effects. 1.3.4 Limits of Detection When looking at a calibration curve, a figur e of merit which should be explained is the limit of detection (LOD). The LOD is the smallest concentration of analyte that can be trusted as present in a sample with a specified level of confidence [Ingle and Crouch 1988]. It should not be confused with the sens itivity, which is the slope of the calibration curve [Kaiser 1978]. The LOD is important in quantifying an analytical technique and is the measurement that expresses advancements in a technique. The LOD is defined by the following relationship: LOD= K /S (1.3)

PAGE 22

10 Where K is the safety factor (most commonly 2 or 3), is the standard deviation of the background signal (taken from a sample contai ning no analyte or the spectral RMS of the noise), and S is the calculated sensitiv ity of the calibration curve (i.e., slope). 1.4 Aerosol Based LIBS Though LIBS is a popular technique for applications in liquids and solids, the gaseous phase, including aerosols, was the ba sis of the following study and is therefore reviewed in detail. An aerosol is defined as solid or liquid par ticles suspended in a gaseous medium. Aerosol particles range from 1 nm to 100 m, with the particles on the lower end of the size distribution (< 2.5 m) receiving more recent research attention due to their increased health implications [U S EPA 1996]. Many techniques based on mass spectrometry principles have been used to determine aerosol characteristics, but more recently LIBS has become a candidate analy tical technique in th e field of individual particle analysis. Due to its small sample size (i.e., plasma volume) in which single particle analysis can be accomplished coupled with statistical calcu lations, studies have demonstrated LIBS as an applicable species mass concentration, and size detector in the laboratory and in the field. 1.4.1 Mass Concentration and Size Detection Radziemski and co-workers throughout th e 1980s developed time-resolved LIBS to overcome high continuum effects at early times and optimize their analyte signal, where they were able to detect chlorine a nd phosphorous aerosols down to 60 and 15 ppm respectively [1981]. Later, Essi en [1988] detected traces of cadmium, lead, and zinc down to detection limits of 0.019, 0.21, and 0.24 g/g, all of which were below the permissible limit reported by the Occupati onal Safety and Health Administration (OSHA). In their study, a liqui d solution with known metal concentration was nebulized

PAGE 23

11 to form an air stream containing a sub-mi cron range of aerosols. Other metal aerosols have been analyzed and detection limits dete rmined, such as As, Be, Cd, Hg, Pb, Sb and Cr [Zhang 1999, Neuhauser 1999]. Also, elements such as F, Cl, and C [Dudragne 1998] along with compounds including phosphine (PH3), arsine (AsH3), and fluorine (CF3H) have been investigated in similar manners [Sneddon and Lee 1999, Peng 1995, and Singh 1997]. Not only have concentration measurements been studied, but also size data has been reported in the liter ature [Hahn 1998]. Hahn and L unden [2000] calculated a size histogram for sub-micron to micron sized ca lcium and magnesium aerosols. Their finding proved that LIBS was a valid technique when sizing aerosols, as their measurements closely corresponded to those by a light scatte ring technique. They determined a lower size LIBS detection limit for the aerosols of 175 nm, and that size measurements could be precisely made even in the presence of ot her aerosol types. Size measurements were made by first determining the concentration of the aerosol by means of comparing the unknown to a known calibration curve. Next, th e absolute mass was calculated using the determined mass concentration and the know n plasma volume. Lastly, by knowing or assuming the composition of the particle and its density, the equiva lent diameter was evaluated. 1.4.2 Continuous Emission LIBS Because of the fast response time possible with LIBS, in situ measurements are one of the biggest advantages in aerosol dete ction, and therefore e xplain much research involving continuous emission studies [Zhang 1999, Neuhauser 1999, Nunez 2000, Ferioli 2003]. Carranza [2001] studied atmosphe ric air, from which concentrations of elements (aluminum, calcium, magnesium, and sodium) where measured over a 6 week

PAGE 24

12 period spanning the 4th of July holiday. Using conditional analysis (methods explained later) they found an order of magnitude increase in alumi num and magnesium during the days adjacent and containing the holiday as a result from the increased fireworks activity. Single shot analysis was also made to meas ure particle diameters, and a histogram was created yielding sub-femtogram mass limits of detection for calcium, and around 2-3 femtograms for magnesium and sodium base d particles. Buckley [2000] performed continuous emission studies of various toxic metals on location at two waste incinerators. Detection limits ranged from 2-100 g/dcsm for Be, Cd, Cr, Hg, and Pb. A conditional analysis was also used to improve detecti on limits over that of more traditional LIBS procedures. 1.4.3 Conditional Analysis Typically ensemble averaging of spectra fr om 100s to 1000s of laser pulses is used to account for the random noise effects from the instrumentation [Radziemski 1994], thus providing a decrease in the spectra l noise. This is useful when the element to be analyzed is constantly inside the plasma volume (e.g., a gas phase element). However, when working with aerosols, especially at concentrat ions near the detection limit, most of the plasmas created are absent of the analyte of interest. This is evident in the statistics offered by Hahn [1997], in which they calculate d sampling rates as low as approximately 0.1 % corresponding to a bout 5 particles/cm3. This means that a total ensemble average would decrease the spectral response (peak-to-ba se and/or signal-to-noise) due to the fact that most of spectra being averaged only contain continuum information and no elemental spectral emission lines. To rectify th ese issues, Hahn [1997 and 2000] proposed a conditional analysis approach that would disc ard the spectra that did not contain atomic information while retaining and averaging spectr a that contained spectra l lines of interest.

PAGE 25

13 Spectra were retained if the ratio of the atomic emission intensity and the adjacent continuum signal exceeded a given threshold. This approach allowed for particle hit rates to be determined as well as improving detec tion limits to the parts-per-trillion. Others have incorporated similar id eas into their res earch including Schechter [1995], Cheng [2000], Martin and Cheng [2000], and Carranza and Hahn [2002a]. 1.4.4 Fundamental Aerosol Studies To understand variations in spectral inte nsity and particle plasma interactions, fundamental single particle an alysis studies have been pe rformed. Specifically, single particle vaporization and their respective signal response as a function of aerosol size has been studied. Cremers and Radziemski [1985] collected beryllium particles on filters allowing for the collection of sizes ranging from 50 nm to 15 m. Spectral information was recorded on beryllium particles throughout the range of sizes. They found that the analyte response of the 15 m particles strayed from the linear increase that occurred for the smaller particles as size was increased. They attributed this to the fact that once the size reached about 10 m, incomplete vapor ization occurred affecting the analyte response. This was the accepted upper-size limit for many years. Carranza and Hahn [2002c] performed a series of experiments in which silica micro-spheres ranging in diameter from 1 m to 5 m were analyzed and single shot spectra were recorded. A linear increase in analyte response was dete cted up to diameters equaling 2.1 m after which the increasing diameters caused the anal yte response to flatte n. They also believed that this was due to incomplete vaporizati on of the aerosol once diameters reached 2.1 m. This study also concluded that partic le vaporization was more dependent on the plasma-particle interactions rath er than a laser-parti cle interaction, due to the fact that the latter interaction would produce sampling rate s much lower than were observed. It was

PAGE 26

14 theorized that some plasma-particle interac tion, such as thermophoretic forces and vapor expulsion, was responsible for the inconsiste ncies found as particle diameters were increased above the upper size limit. Hohreiter [2006] expande d on the theory established by Carranza, in their study which include d a combination of plasma imaging and spectroscopy. Through there methods Hohreiter determined that there is a finite amount of time after the laser pulse when the plasma is energetic enough to disassociate particles (about 15 s for their study). This means that a particle with sufficient size may not be fully disassociated, due to the fact that there is a limited amount of time for disassociation to occur, thus agreeing with the afor ementioned literature by Carranza and Hahn. Other fundamental studies have focused on the forces produced by the expanding plasma volume and the affect they have on the aerosol and gaseous analytes. This is important to understand so that attempts in plasma modeling [Gornushkin 2004] can advance. Historically, it was assumed that atomic emission was independent of the analyte source, and that different sources would generate similar responses as long as equal concentrations of analyte were fully disassociated [Dudragne 1998 and Essien 1988]. For example it was shown that SF6 and HF yielded identic al fluorine signals as long as the mole fractions of fluorine were the same [Tran 2001]. However, in a study by Hohreiter and Hahn [2005a], five equally con centrated different forms of carbon analytes were examined (2 in the solid state and 3 in the gas phase) and were found to have very different spectral emission re sponses despite almost identical plasma characteristics (i.e., temperature and electron dens ity). The solid state analyte had a much stronger response than the three gas carbon anal ytes, which was believed to be attributed to a physical interaction between the analyte and the e xpanding plasma. More specifically, the

PAGE 27

15 expanding plasma was affecting the smaller light er gas phase analyte, forcing it to the perimeter and thereby reducing the amount of analyte presen t within the laser induced plasma, and therefore decreasing the spectral response. Oppositely, th e larger aerosols withstood the force from the plasma expa nsion remaining engulfed in the plasma providing strong emission lines. This physic al plasma-particle phenomenon will be discussed more in later chapters and can be used to explain some of the signal improvements found in the current study. 1.5 Dual-Pulse LIBS Dual-pulse LIBS has been previously inve stigated to a large extent on solid and liquid phase analytes, where it has been de monstrated to significantly enhance atomic emission signal intensity, and more importan tly, to enhance the analyte peak-to-base and signal-to-noise ratios. While the first use of an additional laser pulse to reheat the plasma dates back to the early 1990s [Uebbing 1991], in recent years significant efforts among many research groups have focused on enhancing laser-solid interactions, atomic emission intensity, and detection limits with dual-pulse techniques [St-Onge 1998, Stratis 2000, Gautier 2005]. More specifically, Stratis [2000] used an orthogonal pre-ablation spark to obtain 11 and 33-fold enhancement in spectral response of copper and lead, respectively, in comparison to single laser LI BS. In their study, an orthogonal system was used, in which a laser pulse was brought in parallel to the sample surface and focused a few millimeters above it to form an air plas ma or air spark. A few microseconds later a second laser pulse, which was focused on the sa mple, traveled through the location of the first plasma and ablated the sample materi al, forming the LIBS plasma from which analyte emission was analyzed. Figure 13 illustrates the dua l-pulse orthogonal

PAGE 28

16 configuration along with another setup, which takes advantage of two in-line pulses. Others to find similar enhancement results, some up to 40 times, include First Laser Pulse Focused on the Sample Figure 1-3. Illustrations representing tw o setups used in many of the dual-pulse studies; [A] An orthogonal c onfiguration and [B] An in-line configuration. Sattman [1995], Angel [2001], Coloa [2002] Scaffidi [2004], and Hohreiter [2005b]. Recent studies have examined the mechanisms for this dual-pulse enhancement, focusing on the laser-solid coupling [Hohreiter 2005b, Linder 2005], as well as the hydrodynamics of the overlapping laser pulses, shock waves, and subsequent density effects [Corsi 2004]. The exact mechanisms of signal enha ncement with dual-pulse LIBS are complex due to the combinations of laser ablation, an alyte dissociation, and plasma excitation of atomic species. Previous studies have s hown that enhancements in atomic emission intensity, peak-to-base, and si gnal-to-noise measurements with a dual-pulse system are not simply the result of added energy from th e pre-ablation laser-induced plasma. In fact, dual-pulse signal enhancements are coupled to other physical phenomena such as enhanced ablation, shock-induced plasma rarefaction (i.e., reduced density), or thermal lensing from the pre-ablation plasma [Angel 2001, Coloa 2002, Hohreiter 2005b, Linder

PAGE 29

17 2005]. While much has been learned from these studies of solids and liquids, the extension of dual-pulse LIBS to gas-phase and aerosol system s brings along a new set of dynamics which have not been explored prior to this study. 1.6 Objective Laser induced breakdown spectroscopy is an up and coming quantitative aerosol detection technique, but has problems compe ting with other analytical techniques, especially when comparing detection limits. Th e objectives of this work are to present, test, and compare possible solu tions to increase analyte resp onse in aerosol based LIBS, consequently bettering limits of detection. A dual-pulse system and the use of a particle focusing lens to introduce aerosols into the pl asma were two such methods considered in this work to enhance analyte response. This study is the first to a ddress a dual-pulse laser system for a gaseous and aerosol application, noting that much research has been done on solid and liquid samples as described above. Spectra and plasma transmissi on data from calcium aerosol samples and air samples were collected and analyzed for both the single and dual laser cases. Differences in measurements between the techniques and between the two different samples were noticed, thereby helping to e xplain previously addr essed plasma-particle interactions. The second part of the study focused on th e idea that analyte response could be enhanced when a particle lens was used to introduce aerosols directly into the laser induced plasma. A particle lens aerodynamically focuses aerosol particles so that a tight streamline of particles can be achieved, thus creating a more concentrated cross-section of particles entering the plasma Spectral response and hit rate s of calcium aerosols were compared when they were introduced through a particle lens versus a standard thin

PAGE 30

18 walled tubing. The following study will show that the LIBS technique can still be improved from its current state.

PAGE 31

CHAPTER 2 EXPERIMENTAL METHODS The current study was based on two expe riments, LIBS improvements from a dual-pulse system, and from introducing a partic le focusing apparatus. As in all LIBS configurations, the two experime ntal setups can be divided in to (1) the plasma creation, by means of focusing a laser source so that the threshold of the sample is reached, (2) the collection of the spectral emission from the plasma with the use of a spectrometer and ICCD, and (3) the analyte generation and introduction into the sample chamber. 2.1 LIBS Experimental Setup and Procedure 2.1.1 Plasma Creation Methods High irradiance is needed to cause a brea kdown in a gaseous material. This is done by using a pulsed laser (10 ns duration and 50 mJ/pulse) and by focusing the beam to reduce the cross sectional area. All the beam energy profiles followed a Gaussian trend, peaking at the center of the beam and exponentially decreasing to its minimum at the edges. Figure 2-1 shows a typical setup employed for plasma creation. Nd:YAG laser Gallilean Telesco p e Aperture Condensing Lens Plasma Figure 2-1. Optical setup for the delivery of the laser irradiance to the sample volume resulting in the creation of a plasma. 19

PAGE 32

20 Based on light focusing laws, the diffraction lim it is inversely proportional to the incident beam diameter, meaning that a tighter focu s can be achieved when the beam diameter approaching the condensing lens is large. Fo r this reason, a Gallilean telescope is often used to widen the laser beam prior to its final plasma creating focus. The Gallilean telescope utilizes an expansion lens followed by a lens that collimates the expanding beam. The aperture following the Gallilean telescope trims off the light on the edge of the beam to allow for a more uniform beam profile before it reaches the focusing lens. 2.1.2 Emission Collection Different configurations are used to direct the plasma emission onto the fiber optic before it is processed by the spectrometer. In this study, a back collection method was used. Due to variations in laser pulses and presence of aerosols, the location of the plasma can fluctuate along the beam path ranging up to a millimeter or two. Rather than collecting the light from the side, which w ould result in the fluctuating focused light missing the entrance into the fiber optic, back collection was used so that the error from the spatial variation in the plasma formation is minimized. Once the plasma is created, equally intense light is emitted in every direction with a portion of it being collimated by the same condensing lens used to focus the plasma forming laser. The plasma light was then turned by a pierced mirror, allowing for the laser beam to pass through and the emission to be reflected, and focused by a condensing lens onto a fiber optic bundle le ading into a spectrometer. The light was then dispersed by the 0.275-m spectrometer (Acton SpectraPro -275) and finally recorded by a 256x1024 element intensified CCD array (Princet on Instruments model: 1024MLDS-E/1). The spectrometer used a 2400 groove/mm grating, which provided an approximately 30-nm wavelength range with 0.12 nm spectral resolu tion. The intensifier on a CCD served two

PAGE 33

21 Figure 2-2. Optical setup for the collecti on of plasma emission from the sample volume. Spectrometer ICCD Plasma Emitted Light Fiber Optic Plasma Pierced Mirro r Plasma Creating Laser Radiation Laser purposes, to allow for adjustment of signal ga in and to allow for small time frames in which light is being transmitted (gate width) to the CCD array. The latter is especially important at small delays after the laser pulse, since the continuum light is so intense that it would saturate the CCD camera if small gate widths were not applied. The laser Qswitch is used to trigger a delay generator, wh ich in turn triggers the ICCD, so that light can be acquired at desirable delays after plasma creation. 2.1.3 Aerosol Generation For all aerosol samples, a pneumatic-t ype medical nebulizer (Hudson model 1720) was used to convert an aqueous solution into an aerosol spray. For all experiments, 5 lpm of HEPA filtered air, regulated by a digital fl ow controller (Allicat Scientific), were used to drive the nebulizer. The nebulizer sprayed the aeros ol solution into a secondary flow of air used to dry out the aeroso l droplets and to transport them into the plasma sample volume. Figure 2-3 shows a general schematic for the aerosol generation system. The coflow rates indicated in Figure 2-3 ar e those used in the dual-pulse study.

PAGE 34

22 Mass Flow Controller Co-Flow 20 lpm Mass Flow Controller Co-Flow 20 lpm Mass Flow Controller Nebulizer Flow 5 lpm Nebulizer Porous Plate Plasma Creating Focusing Lens Plasma Sample Volume Drying Chamber Figure 2-3. General schematic of the aerosol generation system. The co-flow rates were those used in the dual -pulse study [Hahn 2001]. Experiments were run to test the analyte concentration consistency in the nebulizer spray. It was possible that what was exiti ng the nebulizer may not be at the same concentration as the solution in the nebulizer due to the selective evaporation of the matrix water. If the spray concentration was not the same as the liquid solution in the nebulizer, then after elapse d running time, the concentration of analyte would be enriched, growing linearly with elapsed time, as compared to the original solution. This was tested by running the nebulizer for 10 and 20 minutes, bo th with a solution containing copper analyte at the same initial concentration, and then comparing what was left over inside the nebulizer to the original solution. The original so lution, along with the

PAGE 35

23 0 500000 1000000 1500000 2000000 2500000 01 02 0 Nubulizer Run Time (min)Analyte Counts Figure 2-4. Chart displaying Cu analyte count s of remnant solutions from a nebulizer as a function of nebulization time. Error bars represent the standard deviation. remnant solutions, were run through an IC P mass spectrometer, from which particle counts of each sample were extracted. Figure 2-4 shows the average particle counts for the original sample as well as the sa mples taken after 10 and 20 minutes. Each measurement was done in triplicate, while each ICP-MS measurement was done in duplicate. Figure 2-4 shows no trend along with very little difference between all three samples, indicating that the concentration in analyte was not changing and therefore the concentration of the aerosol spray exiting the nebulizer was the same as the solution inside the nebulizer over these time scales.

PAGE 36

24 2.2 Dual-Pulse LIBS: Gas and Aerosol Study 2.2.1 Laser Configuration Two Q-switched Nd:YAG lasers operating at their fundamental wavelength of 1064 nm and at a repetition rate of 5 Hz were used in all measurements for this experiment. A schematic of the optical conf iguration is shown in Figure 2-5. The first Figure 2-5. The experimental apparatu s for a single and dual-pulse LIBS configuration. laser (Continuum Precision II 8000), which will be referred to as Laser 1, used a laser beam energy of 100 mJ/pulse, and was focused to the center of the sample chamber (a six-way vacuum cross) using a 100-mm focal length UV grade plano-convex lens. The pulse-to-pulse stability was directly measured (Ophir Nova II), which yielded a relative standard deviation (RSD) of 0.42%, with a maximum pulse deviation of 1.4%. The pulse energy was sufficient to create a laser-induced plasma with Laser 1 operating alone in either a purely gas-phase or aerosol laden sa mple stream. The second laser (Big Sky CFR 400), which will be referred to as Laser 2 operated at a laser pulse energy of 290

PAGE 37

25 mJ/pulse, and was focused with a 75-mm fo cal length UV grade plano-convex lens, also creating a laser-induced breakdow n in the center of the six-wa y cross. The pulse-to-pulse stability was (RSD) of 0.45%, with a maximum pulse deviation of 1.5%. The two laser beams were carefully aligned such that the two plasmas were formed at the identical spatial location (see below fo r alignment details). No quantitative plasma volume measurements were recorded. Howeve r, the Laser 2 plasma was previously characterized in detail; hence a characterist ic plasma diameter is reported as 1.5 mm [Carranza and Hahn 2002b]. Furthermore, the La ser 2 plasma was noticeably larger than the Laser 1 plasma, presumably due to the greater laser pulse energy. As seen in Figure 2-6, the flashlamp sync of Laser 1 was used to trigger a digital delay generator, which was then subsequently used to trigger the flashlamp and Q-switch of Laser 2. Adjustment of the delay generato r allowed for the two laser pulses to fire simultaneously, or at variable pulse-to-pulse delays ranging up to 1 ms, the largest investigated. For all experiments, the delay was adjusted such that Laser 1 was fired first, followed by Laser 2 at the specified delay. The relative flashlamp to Q-switch timing was maintained constant for both lasers for all experiments, thereby ensuring constant laser pulse energy and laser beam characteristics. It is noted that a delay time of zero corresponds to both Laser 1 and Laser 2 pulse s firing simultaneously. A fast response (200-ps rise time) detector and digital os cilloscope (2.5 Gsample/s) were used to continuously measure and monitor the temporal delay between the two laser pulses for all experiments. As implemented, the jitter betw een Laser 1 and Laser 2 was typically less than 5 ns.

PAGE 38

26 Laser 1 Flashlamp Sync Delay Generator 1 To Delay Generator 1 77.75 s + variable delay To Delay Generator 2 To Laser 2 Flashlamp Delay Generator 2A Delay Generator 2B 200 s Fixed To Laser 2 Q-Switch Laser 1 pulse Laser 2 p ulse 277.75 s Fixed Figure 2-6. Schematic representing the tri gger setup to achieve laser pulses fired simultaneously in time or at a variable delay. Delay generator 1 set to 77.75 s would result in the lasers firing simultaneously. Adjusting the variable delay of generator 1 woul d result in delaying laser 2 pulse. 2.2.2 Experimental Analyte 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 [Hahn and Lunden 2000, Hahn 2001]. As seen in Figure 2-3, the purely gaseous sample stream consisted of 40 lpm of purified air, which was passed through an activated alumina dryer, a course particle filte r, an additional desiccant dryer, and finally a HEPA filter cartridge prio r to entering the sample cham ber. All flow rates were controlled with digital mass flow controllers (Matheson model: 8270). For the aerosol measurements, two types of calcium-containing particles were used, with the purpose of studying any potential size effects. The majority of aerosol measurements were made by nebulizing a solu tion of 50 g Ca/ml at a rate of about 0.1 ml/min. The nebulizer output was introduced in a gaseous co-flow stream of 40 lpm of

PAGE 39

27 purified air. The calcium solutions were prepared by diluting ICP-grade analytical standards of 10,000 g Ca/ml (SPEX CertiPrep). Acc ounting for the additional 5 lpm used to drive the nebulizer, th is configuration resulted in a calcium-rich aerosol flow of approximately 100 g Ca/m3 through the LIBS sample chamber. Based on previous TEM measurements using the current configuration [Hahn 2001], the average aerosol particle size following droplet desolvation (i.e., solid an alyte phase) is expected to be less than 100 nm, while agglomerate formation is c onsidered insignific ant. The corresponding particle number density is on the order of 105 cm-3 in the LIBS sample chamber, which yields an average number of analyte part icles per plasma volume on the order of 100. Overall, the system provides a highly dispersed, submicronsized calcium-rich aerosol stream for LIBS analysis. This analyte source will be referred to as the fine calcium aerosol experiments. In addition, some experiments were perf ormed by nebulizing a suspension of nominally 2m sized ( 0.7 m) borosilicate glass micros pheres (Duke Scientific, #9002) in ultrapurified water. Based on prev ious analysis, the calcium concentration within the glass microspheres was determined to be about 2% (by mass), which yields a strong calcium atomic emission signal [Hohre iter 2006]. The particle concentration in suspension was adjusted so that the resu lting borosilicate par ticle number density produced in the LIBS sample chamber was on the order of 102 cm-3. These experimental conditions will be referred to as the borosilicate microsphere experiments. 2.2.3 Spectral Measurements Both laser-induced plasmas were first visua lly aligned to the same spot, noting that the Laser 2 plasma emission was previously a ligned to the fiber optic such that both

PAGE 40

28 atomic and continuum plasma emission were optimized. Once the Laser 1 plasma was visually centered on the Laser 2 plasma, the final alignment of Laser 1 was performed to maximize plasma emission (atomic and continuu m emission) coupling to the fiber optic, thereby ensuring alignment of both laser-induced plasmas to the same spatial location. For all single-pulse and dual-pulse experiments, the external Q-switch sync from Laser 2 was used to trigger the ICCD controlle r. Hence for a given set of experiments, the ICCD was fixed relative to the temporal posit ion of Laser 2. An additional delay was then introduced between Laser 2 and the ICCD detector gate, which allowed for optimization of the specific analyte atomic emission si gnals. For the purely gas-phase experiments (i.e., nitrogen and oxygen atomic emission analys is), an ICCD gate delay and gate width of 5 s were used, therefore spectral integration was initiated 5 s following Laser 2. For the aerosol experiments (i.e., calcium atomic emission analysis), both the detector delay and the gate width were increased to 30 s. Spectral data was acquired using an ense mble average of 1000 laser shots. The process was repeated a total of three time s for each different analyte, and for each adjusted dual-pulse delay. In addition to dual-pulse measurements, single-pulse data were recorded for both Laser 1 alone and Laser 2 alone for all different analytes. Finally, single-particle analysis was used for the 2m sized borosilicate glass particles, as reported in previous studies [Hahn and Lunden 2000, Hohreiter 2006]. These single-shot experiments were performed using 500 shot sequences, repeated three times for each experimental condition.

PAGE 41

29 2.2.4 Transmission Measurements For the transmission experiments, laser pul se energy measurements were made for Laser 2 at a spatial location dire ctly in front of the sample chamber (i.e., incident energy) and at a location directly exiting the sample chamber (i.e., transmitted energy). This was achieved by placing the laser energy meter (Ophi r Nova II) in front of the six-way cross to measure the incident beam energy, and im mediately after the chamber to measure the transmitted laser beam energy, with the latt er configuration shown in Figure 2-5. The average transmitted energy (i.e., Laser 2 transmission) was then calculated from the direct ratio of these two measurements. The transmission measurements were record ed for both single-pulse (Laser 2 only) and dual-pulse (Laser 2 following Laser 1) configurations, for both the gaseous and aerosol systems, including a full range of dual-pulse, laser-laser delay times. An ensemble-average of 500 laser shots was recorded for both the incident and transmitted pulse energies, which were repeated a mini mum of three times each. In addition to the mean energy values, full statistical para meters were recorded, which included the minimum, maximum, and standard deviati ons of the shot-to-shot pulse energies. 2.3 Particle Lens Study 2.3.1 Laser Configuration An Nd:YAG laser (Continuum Surelite II) w ith a repetition rate of 5 Hz was run through two wavelength doubling crystals yi elding a 355 nm plasma creating laser source. As seen in Figure 2-7, the laser source was turned by a series of 355 nm dichroic mirrors with anti-reflection coatings at 532 and 1064 nm to eliminate any excess fundamental 1064 nm and green 532 nm light. The beam energy meas ured after the 355 nm mirrors just before reaching the condens ing lens was about 60 mJ/pulse. The laser

PAGE 42

30 beam passed through a pierced mirror and wa s finally focused by a 100 mm focal length UV grade plano-convex lens to create the plasma. Aerosol and Air Mixture Vacuum Pumped Out Nd:YAG laser 355 nm AR mirrors Particle Lens Pierced Mirror Fiber O p tic Figure 2-7. The optical setup (top view) for the particle lens experiments. Like the dual-pulse setup, the plasma emission was back collected. The aerosols were drawn through the particle lens into the plasma by a vacuum pump situated on the opposite side of the 5-way cross. 2.4.2 Analyte Generation For all measurements, the aerosols were created by nebulizing a solution of nominally 2m sized ( 0.7 m) borosilicate glass micros pheres (Duke Scientific, #9002) in ultrapurified water, th e same used for the borosilic ate experiments in the dualpulse study. Accounting for 5 lpm used to driv e the nebulizer and the 5 lpm of co-flow used to dry the nebulized droplets, a flow was created yielding approximately 30 g Ca/m3. A portion of the nebulized flow, about 1 lpm (based on the optimal conditions determined by the manufacturer, Aerodyne Research, Inc.) was pulled through the particle lens and the plasma chamber by a vacuum pump. The flow rate through the vacuum pump was regulated by a rotameter-t ype flowmeter (Gilmont Instruments). To

PAGE 43

31 determine the particle lens effectiveness, a st andard thin walled steel tube with an inner diameter of about 0.18 was also used in all experiments to transpor t the aerosols to the plasma. The particle lens was developed by Aer odyne Research, Inc. and consisted of a series of concentric tubes that stepwise na rrowed the flow down to a cross section with a diameter of 1.65 mm before re leasing the streamlined part icle flow out of a 0.75 mm orifice port. Figure 2-8 s hows a general schematic of the particle lens. Particle Lens Figure 2-8. Simple schematic of the particle lens. Aerosols flow through a series of radial decreasing concentr ic tubes before exiting through an outlet causing a streamline exit into th e laser induced plasma The particle lens outlet was visually aligne d vertically and horiz ontally to the laser induced plasma and was situated about 2 mm away. At first attempts, it was determined that the particles were filling and re-circulating throughout the entire sample chamber (the 5 way cross), therefore rendering the particle lens useless, since there was no preferential increase in particle concentrati on at the exit of the particle lens. To remedy the recirculation problem, a secondary shea th flow was introduced annularly around the particle lens. This was accomplished by placi ng a thin walled steel tube around the particle lens and the tube, represented by the outer lines in Figur e 2-8. Air was flown

PAGE 44

32 through the tube at a flow rate of 9 lpm and was shown to be effective in separating the particles exiting the lens with those re-circulating in the chamber. 2.4.3 Spectral Measurements Similar to the other study, the resulting emission was back collected, turned by the pierced mirror and focused by a 100 mm focal le ngth lens onto a fiber optic bundle. The fiber optics fed into the same spectrometer and ICCD that was used in the spectral collection for the previous study. The ICCD controller was triggered by the Nd:YAG lasers external Q-switch, initiated 20 s after the laser pulse with a 30 s gate width. Analyte response was compared using spect ral data acquired through an ensemble average of 1000 laser shots for each type of aer osol injection tube (the particle lens and the thin walled tube). A conditional anal ysis approach, mentioned in Chapter 1, was also used to gather hit rates and to acquire individual spectra that could be used to further investigate any advantages in using a particle lens. 2.4.4 Particle Counting Measurements To help quantify the difference in hit ra tes and analyte respon se between the two injection methods, the transmission of partic les through both tubes was calculated. This was done by flowing the analyte through the t ube and the particle lens both at 1 lpm. A particle counter was placed at the end of the 5-way sample chamber, which drew out 0.28 lpm from the exiting 10.28 lp m flow (combination of the sheath flow, 9.28 lpm, and the analyte co-flow mixture, 1 lpm). The part icle counter would calculate the number of particles between a certain si ze range. Since the experiment al analyte nominal diameter was 2.2 m, all counts of particles above 2 m were considered. Th e particle counter used was a Lasair model 1001 (Particle Measuri ng Systems Inc.) that used light scattering to determine particle number and size.

PAGE 45

CHAPTER 3 RESULTS AND DISCUSSION 3.1 Dual-Pulse LIBS Study This study focused on the effects of an orthogonal dual-pulse laser configuration on the atomic emission response for both purely ga seous and calcium-based aerosol samples. The purpose was to determine if a dual-pulse configuration, which has been shown to increase analyte response in solids and liquids, could improve a gas and aerosol analyte response over that of a typical single-pulse LIBS system. Transmission measurements were made to help physically explain the differences observed in analyte response between the two configurations and be tween the gas and aerosol analyte. 3.1.1 Transmission Experiments Figure 3-1 presents the transmission of Laser 2 as a function of delay time between Laser 1 and Laser 2 for both the pure air stream and the fine calcium aerosol experiments. As a reference, the dashed line in Figure 3-1 represents the average transmission of 43.8% for the single-pulse configuration (i.e., Laser 2 only), which is the average of the purified air (44.5% with 2.7% RSD) and the fine calcium partic le (43.1% with 0.8% RSD) transmission values. The overall behavior of the transmission data as observed in Figure 3-1 is rather complex, and is consider ed for discussion purposes in terms of four distinct temporal regions. The average RS D values for the dual-pulse transmission measurements were 4.4% for the pure air str eam and 3.3% for the fine calcium particles, as averaged over all temporal delays. The corresponding error bars where comparable to 33

PAGE 46

34 the symbol size for most values, hence error bars were omitted in Figure 3-1 to avoid clutter, given the partial overlap of many symbols. De 102103104105106 0.0 0.2 0.4 0.6 0.8 1.0 020406080100 Pure Air Calcium AerosolTransmissionDelay Time (us) Delay Time (ns) Figure 3-1. Laser 2 transmission as a functi on of dual-pulse laser delay times for the pure air and fine calcium aerosol sample streams. The horizontal line represents the transmission of Laser 2 alone (i.e., single pulse LIBS). Note the plot is linear for delay times less than 100 ns. The first temporal region to be consid ered will correspond to dual-pulse delays less than 100 ns. During this temporal range, there was litt le transmission of Laser 2 through the plasma formed by Laser 1, as explained below. The minimum recorded transmission was 3.3% (1.3% RSD) and 3.1% (8.2% RSD), which both occurred at a delay of 25 ns, for the gas-phase and fine cal cium aerosol phase, respectively. At these

PAGE 47

35 delays, the plasma formed by Laser 1 is esse ntially opaque to the in cident Laser 2 pulse, which follows a previously reported trend for the temporal characteristics of laserinduced plasmas [Hohreiter 2004b]. During th is temporal region, the high plasma temperatures and free electron densities (~1018 cm-3) of the first plasma result in an optically dense plasma for the Laser 2 incide nt radiation. The resulting highly energetic plasma state is essentially independent of the presence of the calcium aerosol, therefore no difference is observed between the two experimental conditions. This result is consistent with previous measurements, in which identical plasma temperatures and free electron densities were recorded for gas-pha se and particle-seeded flows under conditions similar to the present study [Hohreiter 2005a], and no effect of aerosol presence on the temporal location of plasma initiation was r ecorded [Hohreiter 2004a]. Finally, it is noted that qualitative observations of the plasma revealed no differences between the pure air and the aerosol seeded conditions, with regard to plasma size and spatial stability. The second temporal region of interest in Figure 3-1 is considered to range from a delay of 100 ns to about 1 s. During this region, the highly absorptive nature of the first plasma decreases as the free electron dens ity and temperature rapidly decrease. Once again, this is consistent with previous measurements, where an essentially identical laserinduced plasma was found to be nearly tran sparent to a low-ener gy probe beam by about 500 ns following plasma initiation [Hohreiter et al 2004b]. Overall, as the delay between Laser 1 and Laser 2 nears about 1 s, the effect of the Lase r 1 plasma is not very significant regarding the coupling of Laser 2 energy into the existing Laser 1 plasma, as the overall transmission of Laser 2 is near its si ngle-pulse value. In other words, with the dual-pulse configuration, Laser 2 initiates a breakdown process comparable to what

PAGE 48

36 would occur in the presence of Laser 2 al one, resulting in a similar coupling of the incident energy into the resulting plasma. The third temporal region corresp onds to delays ranging from 1 s to about 100 s, and is defined by a relatively high transm ission of the Laser 2 pulse. Specifically, the transmission of Laser 2 has a maximum value of 90.0% (0.82% RSD) and 79.5% (1.1% RSD) for the pure air and fine calcium aerosol systems, respectively, over this temporal region. Unlike in the earlier temporal regions, an effect of the fine cal cium aerosol is now observed, as the transmission is slightly reduc ed in the presence of aerosol in comparison to the pure gaseous system. This reduction is indicative of some laser-particle interactions during the breakdown event, as di scussed in more detail below. Finally, the fourth temporal region co rresponds to laser-laser delays from 100 s up to 1 ms. At the latter value, the transm ission of the Laser 2 beam is observed to approach the average single-pulse (i.e., Laser 2 only) value. Clearly at such large delays, the Laser 1 plasma has sufficiently decayed in both temperature and free electron density such that its effects on the subsequent la ser pulse (i.e., Laser 2) are negligible. The transmission behavior of the first two regions is not unexpected, given the earlier studies on temporal plas ma characteristics; hence a st rong plasma-laser interaction during this period drives the dual-laser coupling and is virtually independent of the presence of particulates. Furthermore, the beha vior of the last region is expected as well, given that at significantly long laser-laser delays, the two laser pulses must approach independence with regard to interaction. Clearly ~1 ms is nearing this asymptotic limit for pulse-to-pulse independence. Given these comments, a key region of interest with regard to the physics of dual-pulse LIBS fo r gaseous and aerosol systems is the third

PAGE 49

37 region, namely between about 1 to 100 s of laser-laser delay. It is well known that the laser-induced plasma is characterized by a rapid plasma expansion and a concomitant shock wave. Not withstanding the increased pressure behind the shock front, the significant plasma temperature results in a re duced density (i.e., mass/volume) within the resulting plasma. This rarefaction is important with regard to the coupling of Laser 2 into the existing Laser 1 plasma, as re lated to the breakdown threshold. At ambient temperature, the laser-induced breakdown threshold is known to vary inversely as pressure, ideally as p-1/2, noting that both multiphoton and cascade ionization processes are important in plasma formation and growth. Therefore, the ionized plasma from Laser 1, albeit at reduced density, presen ts a more complex prob lem for treatment of the Laser 2 interaction than w ould be predicted from treatment of pressure/density effects alone. However, the significant decrease in Laser 2 energy coupling into the plasma (i.e., increased transmission) is interpreted in te rms of an increased br eakdown threshold. This increased threshold effectively delays the temporal breakdown point of the Laser 2 pulse to later in the pulse waveform, there by resulting in less coup ling of Laser 2 pulse energy. It is well known that the presence of aer osol particles can c onsiderably lower the breakdown threshold [Lencioni 1972 and Smith 1997]. Therefore, the slight reduction in transmission between 5 and 20 s delay from the pure air stream ( = 89.7% with 0.7% RSD) as compared to the fine aerosol stream ( = 75.0% with 1.5% RSD) is consistent with the addition of the calcium-rich pa rticulates and a concomitant reduction in breakdown threshold.

PAGE 50

38 3.1.2 Spectral Analysis of Gaseous Analyte Neutral atomic nitrogen (N I) li nes at 491.4 nm and 493.5 nm (86,137 106,178 cm-1 and 86,221 106,478 cm-1, respectively), and the oxygen triplet centered at 394.7 nm (73,768 99,094 cm-1) were used for the spectral analysis of the purified air sample stream. As an example, nitrogen spectra for the single-pulse (Laser 2 only) and the dualpulse (500 ns delay) configuratio ns are presented in Figure 3-2. 0 2000 4000 6000 8000 10000 12000 4 8 04 8 54 9 04 9 5Intensity (a.u.)Wavelength (nm) Laser 2 Only Dual Pulse 5 0 0 Figure 3-2. Spectra showing the two nitrog en atomic emission lines at 491.4 nm and 493.5 nm. The spectra correspond to dua l-pulse LIBS with 500 ns delay (lower spectrum) and single-pulse (L aser 2 only) LIBS (upper spectrum). Both spectra have the same scale. Recalling that the ICCD is also synchronized to Laser 2, it is observed that the overall signal intensity is somewhat reduced by the dual-pulse scheme for this temporal

PAGE 51

39 regime. However, at significantly short laser-laser delay times (~0 to 100 ns), the dualpulse configuration produced a slightly greater emission in tensity than Laser 2 alone, which is consistent with the enhanced coupling efficiency of Laser 2 as observed in the transmission experiments. To quantify the emission signals, both th e peak-to-base and signal-to-noise ratios were calculated using the 493.5-nm N I spect ral line for both the dual-pulse and the single-pulse configurations. The peak-to-base is perhap s the most widely used LIBS signal metric, as it provides a normalization of the atomic emission line with the plasma continuum emission, allowing for a more preci se emission signal as noted previously [Coloa 2002 and Carranza and Hahn 2002a]. As an analytical figure of merit, the signalto-noise ratio is the more relevant metr ic. The signal-to-noise ratios (SNR) were calculated from the integrated peak intensities and the calculated root-mean-square noise from the adjacent continuum region. Figures 33 and 3-4 present the peak-to-base ratio (P/B) and signal-to-noise ratio as a function of dual-pulse delay for the 493.5-nm nitrogen line. Before discussing the figures, it is impor tant to note that La ser 1 alone produces a significant laser-induced plasma; hence anal yte emission is observed over a range of detector gate delays stemming from this em ission source only. Therefore, as the laserlaser delay is being adjusted for dual-pulse ex periments, the effective detector gate delay with respect to Laser 1 is also being varied, noting that the detector gate is fixed with respect to Laser 2. It is therefore important to consider the analyt e signals stemming from Laser 1 alone (i.e., Laser 2 blocked), which ar e therefore included in Figures 3-3 and 3-4.

PAGE 52

40 18 20 22 24 26 28 02004006008001000 Laser 1 Only Dual Pulse Cont. 490 DP 490Peak-to-Base Ratio (493.5-nm N)Delay Time (ns) Figure 3-3. The 493.5-nm nitrogen emission lin e peak-to-base ratio measurements for the pure air sample as a function of dual-pulse laser delay times. The P/B ratios are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizon tal line represents the average P/B ratio for Laser 2 only. A representative error bar is included on the 100 ns Laser 1 only mark. Figure 3-3 shows the dual-pulse results as compared to the single-pulse LIBS configuration (i.e., Laser 2 only), which is in dicated with the dashed horizontal line. For the range of delays from 0 to 1000 ns, the P/B ra tios are observed to depart little from the single-pulse average of 22.8 (0.21% RSD). The maximum dual-pulse value occurs at 250 ns delay, and corresponds to a P/B ratio of 23.8 (0.41% RSD), or a 4% increase with the dual-pulse configuration. While statistically significant, a 4% signal enhancement would not justify the increased expe rimental complexity and expense of the dual-pulse scheme. By 1 s delay time, the dual-pulse P/B ratio has decreased to 21.3 (0.47% RSD), or a

PAGE 53

41 reduction of 6.8% with regard to single-pulse LIBS. Data were collected for larger laserlaser delay times, but this trend of decreas ing response continued, with a dual-pulse P/B ratio of only 9.3 (0.50% RSD) recorded at 5 s delay. This corresponds to a 60% decrease in analyte signal with the dual-pulse scheme at this larger delay, as a result of the diminished Laser 2 coupli ng consistent with Figure 3-1. 0 10 20 30 40 50 60 70 02004006008001000 Laser 1 Only Dual PulseSignal-to-Noise Ratio (493.5-nm N)Delay Time (ns) Figure 3-4. The 493.5-nm nitrogen emission lin e signal-to-noise ratio measurements for the pure air sample as a function of dual-pulse laser delay times. The SNR are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizon tal line represents the average SNR ratio for Laser 2 only. The SNR values, as shown in Figure 3-4, display a similar trend to the P/B ratios, although the maximum enhancement for the dual-pulse configuration is shifted from 250 ns delay to a delay of 500 ns. Specifically, th e SNR is increased from a single-pulse value

PAGE 54

42 of 33.0 (0.47% RSD) to a dual-pulse valu e of 47.6 (2.4% RSD) at 500 ns, which corresponds to a 44% impr ovement. Once again, by a 5 s laser-laser delay, the dualpulse SNR was decreased by 86%, to a value of only 4.6 (0.10% RSD). The oxygen atomic emission lines reveal ed similar, although not as positive, trends as the nitrogen data. Figure 3-5 displays the 394.7 nm oxygen spectral emission line for the 500 ns laser-laser delay dual-pulse case as well as the sp ectra for the single laser 2 only case. The P/B ratios were calcu lated using the 394.7-nm O I triplet and can 0 2000 4000 6000 8000 10000 376380384388392396400404Intensity (a.u.)Wavelength (nm) Laser 2 Only Dual-Pulse Figure 3-5. Spectra showing the oxygen (I) triplet at 394.7 nm. The spectra corresponds to dual-pulse LIBS with 500 ns delay and single-pulse (Laser 2 only) LIBS. Both spectra have the same scale.

PAGE 55

43 be seen in Figure 3-6. For a laser-laser delay of 250 ns, the P/B ratio was decreased from a single-pulse value of 18.1 (0.79% RSD) to a dual-pulse value of 17.0 (0.40% RSD), or a decrease of 6.5%. By 5 s, the dual-pulse P/B ratio was reduced to 4.2 (8.1% RSD), corresponding to a reduction of 76.9%. A signal-to-noise plot was also constructed for the 394.7 nm oxygen spectral emission line, whic h can be seen in Figure 3-7. The trend of the S/N plot followed similarly to the oxyge n P/B plot. Like the P/B plot, the signal-tonoise ratios indicate a decrease in analyte response for the du al-pulse configuration when compared to the single Laser 2 only (r epresented by the dashed line) case. 12 13 14 15 16 17 18 19 20 02004006008001000 Laser 1 Only Dual PulsePeak-to-Base Ratio (394.7-nm O)Delay Time (ns) Figure 3-6. Oxygen peak-to-base measurem ents for the filtered air sample as a function of dual-pulse laser delay tim es. The average peak-to-base ratio for Laser 2 only is represented by the dashed horizontal line.

PAGE 56

44 10 15 20 25 30 35 40 45 02004006008001000 Dual-Pulse S/N Laser 1 Only S/NSignal-to-Noise Ratio (394.7 nm O)Delay Time (ns) Figure 3-7. The 397.4-nm oxygen emission line signal-to-noise ratio measurements for the pure air sample as a function of dual-pulse laser delay times. The S/N are also shown corresponding to th e Laser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizonta l line represents the average S/N ratio for Laser 2 only. The above results appear at first glance to contrast with much of the current literature on dual-pulse LIBS, which tends to show significant signal enhancements with dual-pulse schemes. However, the mechanis ms and physics of purely gas-phase LIBS analysis are fundamentally different from the analysis of bulk solids or liquids. With the latter, the reduced density behind the shock wave of the first plasma enhances the second lasers interaction with the solid sample, while simultaneously reducing the plasma coupling with the gaseous matrix; thereby increasing the analyte response of the solid. In contrast, the present results must be expl ained in the context of the transmission

PAGE 57

45 experiments, as well as in terms of recen t experiments compari ng the response of gasphase and particulate phase analytes [Hohreiter 2005a]. For the relatively short laser-laser delay times ( ~ 1 s), the Laser 2 pulse is efficiently coupled to the Laser 1 plasma, leading to a greater quantity of energy coupled into the resulting dual-pulse plasma. Increased total lase r pulse energy into a laserinduced plasma event may or may not improve th e analyte signal respon se, as explored in detail in an earlier study [Carranza and Hahn 2002a], depending on the overall plasma regime. In the current study, the additional energy with the dual-pulse configuration produced a marginal increase in nitrogen em ission, and a marginal decrease in oxygen emission, during this delay period. This trend most likely reflects th e slightly different upper energy states of the nitrogen and oxygen transitions, which are expected to correspond to slightly different optimal temporal windows, as based on plasma temperature [Fisher 2001]. By coupling more en ergy into the resulti ng dual-pulse plasma, the temporal temperature profile of the plas ma is altered, thereby altering the peak-tocontinuum emission characteristics at the fixed detector gate. This behavior is further reflected by comparing the Laser 1 only data to the Laser 2 only data at zero delay time, which has the effect of changing the laser pulse energy from 100 mJ to 290 mJ with a fixed detector gate. The nitrogen P/B ratio is observed to increase by 3% with increased pulse energy, while the oxygen P/B ratio is observed to decrease by 5%. Clearly the temporal location of optimal P/B response is different for the selected nitrogen and oxygen emission lines. Notwithstanding the above comments, for the longer laser-laser delay times (beyond 1 s), a very different dynamic is obser ved for both gaseous analyte species.

PAGE 58

46 Specifically, for this increased delay regi me, the dual-pulse configuration yielded a significant decrease in analyte response, which again is explained in the context of Figure 3-1, and the generally lessened LIBS analyte response with gas-phase species as recently reported [Hohreiter and Hahn 2005a]. In thei r study, Hohreiter and Hahn proposed that the expanding shockwave of the laser-induced plasma would preferentially expel (i.e., push) molecules from within the plasma core, producing a decreased concentration of analyte for gas-phase species, represented in Figure 3-8. Figure 3-8. Image detailing the proposed mechanism, in which an expanding plasma volume will force the nano-scale gas phase particles outward depleting their concentration in the plasma core [Hohreiter and Hahn 2005a]. In addition to this depletion of analyte, the significantly reduced coupling of Laser 2 into the existing Laser 1 plasma (i.e., enhan ced transmission per Figure 3-1) effectively negates the dual-pulse advantage. This effect results in an an alyte signal that, for the most part, corresponds to the Laser 1 plasma alone, as seen in th e excellent agreement between the dual-pulse and Laser 1 only data in Fi gures 3-3 and 3-4, as well as for the oxygen emission data. Stated another way, the dual-pu lse advantage of de-c oupling the gas-phase matrix from the bulk analyte phase, as reali zed with solid and liquid analysis, is not

PAGE 59

47 possible with a dual-pulse analysis of a purely gas-phase system, because the gas-phase matrix itself is the actual analyte of interest. 3.1.3 Spectral Analysis of Fine Calcium-Based Aerosol Analyte In contrast to the above gasphase analyte results, the anal ysis of the aerosol sample streams affords an opportunity to realize the benefits of dual-pulse LIBS by attempting to 0 500 1000 1500 2000 2500 385 390 395 400Intensity (a. u.)Wavelength (nm) Dual Pulse Laser 2 Only Figure 3-9. Spectra showing the Ca II atomic emission lines at 393.4 and 396.9 nm for both the dual-pulse configuration with a 250-ns delay, and for Laser 2 only. Both spectra have the same intensity scale, and the dual-pulse spectrum has been shifted upw ard by 400 counts for clarity. decouple the particulate-phase derived analyt e from the gas-phase species. With the addition of aerosol particles into the sample stream, distinctly different results as

PAGE 60

48 compared to the gas-phase experiments were observed. For these experiments, the first ionized calcium (Ca II) atomic emission lines at 393.4 and 396.9 nm (0,414 cm-1 and 0,192 cm-1, respectively) were used for all spectral measurements due to their strong intensity. As an example, reco rded spectra are presented in Figures 3-9, 3-10, and 3-11, 0 500 1000 1500 2000 2500 385 390 395 400Intensity (a.u.)Wavelength (nm)Dual Pulse Laser 2 Only Figure 3-10. Spectra showing the Ca II at omic emission lines at 393.4 and 396.9 nm for both the dual-pulse configuration with a 750-ns delay, and for Laser 2 only. Both spectra have the same intensity scale, and the dual-pulse spectrum has been shifted upw ard by 400 counts for clarity. as acquired using the dual-pulse configurati on with a laser-laser delay of 250 ns, 750 ns, and 50 s, respectively. Along with the dualpulse spectra, for comparison, spectra from the single-pulse (Laser 2 only) configuration is also presented in each of the mentioned figures.Several features are noted in Figure 310, which represents th e 750 ns laser-laser delay. The relative intensity of the Ca II atomic emission peaks is significantly greater for

PAGE 61

49 the dual-pulse method as compared to the single-pulse method for, namely 1980 counts vs. 1300 counts for the 393.4 nm peak, respectivel y. In addition, the region of the spectra to the left of 393.4-nm Ca II line corr esponds to molecular emission from the first negative system, including the lines at 391.4 an d 388.4 nm. In contrast to the particle2N 0 500 1000 1500 385 390 395 400Intensity (a. u.)Wavelength (nm) Dual Pulse Laser 2 Only Figure 3-11. Spectra showing the Ca II at omic emission lines at 393.4 and 396.9 nm for both the dual-pulse configuration with a 50-s delay, and for Laser 2 only. Both spectra have the same intensity scale. derived calcium emission lines, these gas-pha se derived molecular lines display an opposite trend, whereby the emission intensitie s are significantly re duced with the dualpulse configuration. This latter trend is perfectly consistent w ith the above analysis of the gas-phase atomic nitrogen and oxygen emission data.

PAGE 62

50 In a similar manner as for the gas-phase experiments (noting the longer detector gate and width of 30 s), comparisons of Ca II atomic emission peak-to-base and signalto-noise ratios were made for the single-p ulse and dual-pulse conf igurations, with the results presented in Figures 3-12 and 3-13, respectively. 103104105 0 100 200 300 400 500 600 02004006008001000 Laser 1 Only Dual Pulse Peak-to-Base Ratio (393.4-nm Ca)Delay Time (ns) Figure 3-12. The 393.4 -nm calcium II emission line peak-to-base ratio measurements for the fine calcium aerosol sample as a function of dual-pulse laser delay times. The P/B are also shown corre sponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked). The da shed horizontal line represents the average P/B ratio for Laser 2 only. The dual-pulse measurements for both th e peak-to-base and signal-to-noise are significantly greater than those realized w ith the single-pulse scheme over a range of laser-laser delay times from about 100 ns to nearly 100 s. Specifically, the P/B ratio for the single-pulse experiments (Laser 2 only) was an average of 220 (7.0% RSD). With the

PAGE 63

51 dual-pulse configuration, the P/B was observed to increase to values of 460 (6.7% RSD) at 750 ns delay, and 463 (4.6% RSD) at a delay of 5 s. Nearly identical trends were observed with the SNR data, as seen in Fi gure 3-13. The average si ngle-pulse (Laser 2 only) SNR value was 43.4 (7.1% RSD), which was increased to a maximum value of 103104105 0 50 100 150 200 250 02004006008001000 Laser 1 Only Dual Pulse Signal-to-Noise Ratio (393.4-nm Ca) Delay Time (ns) Figure 3-13. The 393.4 -nm calcium II em ission line signal -to-noise ratio measurements for the fine calcium aerosol sample as a function of dualpulse laser delay times. The SNR are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizontal line represents the average SNR for Laser 2 only. 202 (10.5% RSD) for a dual-pulse configur ation with a laserlaser delay of 5 s. This corresponds to a greater than 4-fold increase in analyte sensitivity with the optimal dual

PAGE 64

52 pulse scheme, which corresponds to a laser-la ser delay in the range of about 0.8 to 5 s. It should also be noted that the Laser 1 (s ingle laser only) P/B and SNR data are all markedly less than the dual-pulse results. The Laser 1 response corroborates the dualpulse enhancement, and results from a stea dy decrease in calcium emission response with increasing laser-to-detector delay time, as effectively realized with the Laser 1 only experiments. Recall that with Laser 1 only, th e detector gate is still being temporally delayed even though the Laser 2 beam is blocked. As discussed above with the purely gas-pha se data, the aerosol-phase experiments should also be explained in the context of the transmission experiments and the recent gas-phase vs. particulate-phase analyte response study [Hohreiter and Hahn 2005a]. Referring once again to this earlier study, th e solid particulates (i.e., aerosol phase analyte) were hypothesized to resist being driven from the plasma center along with the gaseous molecules by the expanding shock wave as seen in Figure 3-14. The resulting Figure 3-14. This image depicts the basis for the increase analyte response in a dualpulse configuration for an aerosol sa mple. The larger aerosol particles resist the radial force exerted on them by the expanding plasma volume created by laser 1 and remain inside the core while the smaller gas phase particle are expelled, th ereby creating a more aerosol concentrated volume that is awaiting the plasma from laser 2 [Hohreiter and Hahn 2005a].

PAGE 65

53 slip factor has the effect of preferentially enhancing the particulate-phase/gas-phase analyte ratio, thereby affordi ng an opportunity to enhance the analyte response of the solid phase with an optimally-timed second laser pulse. In other words, the gas-phase species within the plasma effectively contri bute to the plasma continuum emission, hence to spectral noise with regard to the targ eted particulate-derived analyte emission. In keeping with the Figure 3-1 transmission data the temporal region between about 1 and 10 s is consistent with a proposed mechan ism in which the particulate-phase is preferentially enriched by the loss of gas-pha se species, and is then additionally excited (i.e., strong coupling) by the second laser pul se (Laser 2). At relatively shorter delay times (< 500 ns), the dual-pulse enhancement is not as great because the rarefaction has not yet developed, although plasma-laser coup ling is still strong. At relatively longer delay times (>10-20 s), the first plasma has undergone substantial decay, thereby reducing the coupling of Laser 2, and the sy stem is once again approaching conditions corresponding to a single-pulse environment, as seen in the spectra of Figure 3-11. 3.1.4 Spectral Analysis of Boro silicate Glass Microspheres To further examine the observed phe nomena of dual-pulse enhancement with aerosol analysis, in the context of the mech anisms offered above, additional experiments were performed using significantly larg er aerosol particles, namely the 2m borosilicate microspheres. This was done in an effort to determine if the plasma -particle interactions with the dual-pulse configuration were furthe r enhanced with increasing particle size. Based on recent work with the identical part icles, the overall time scale for particle dissociation within the laser-induced plasma was estimated to be on the order of 15 s

PAGE 66

54 [Carranza and Hahn 2002a], hence particulates are expected to be present during the currently determined optimal dual-pulse lase r-laser delays of between about 1 and 10 s. For these experiments, suspensions of the borosilicate glass were nebulized as described above, and single-shot spectra were analyzed for the presence of calcium atomic emission using the same 393.4-nm Ca II atomic emission line. Measurements were made using a single-pulse configuration (Laser 2 only) and a dual-pulse configuration with a laser-las er delay of 250 ns. While this delay was not the optimal value per the SNR experiments, the single-s hot detection criteria makes use of the P/B ratio [Carranza 2003]; hence the 250 ns was close to the optimal value per Figure 3-12, and had the added advantage of slightly larger absolute signal counts to work with. Because the particle hit rate was adjusted (i.e., aerosol number density reduced) to be less than 100% to minimize multiple pa rticle sampling with a single shot, any ensemble-averaging would reduce the effect of dual-pulse enhancement on calcium emission by averaging the calcium-containing spectra (i.e., partic le hits) with noncalcium containing spectra (i.e., particle free sp ectra ). Normally, this might be addressed by using a conditional processing routine to separate out the spectra corresponding to particle hits, and to then analyze only such spectra [Hahn 1997]. However, the easy solution of identifying and ensemble-averaging th e spectra of calcium-based particle hits brings an additional problem. By enhancing th e sensitivity of calcium detection with the dual-pulse configuration, the particle hit rate is expected to increase. However, since particle hits co rresponding to the strongest calcium-emission signals are most likely to be sampled with both single-pulse and dual-pulse configurations the gains in sampling rate with dual-pulse LIBS are expect ed to be made for the par ticle hits corre sponding to the

PAGE 67

55 weakest calcium emission signals. Therefore, comparing an ensemble-average of a larger number of spectra that contain a larger percentage of weak er emission signals, to an average containing a smaller number of spectra but with stronger emission signals is not a valid comparison. Because development of a detailed algorithm to attempt to sort and categorize spectra according to emission signa l distributions was beyond the scope of the present study, the borosilicate pa rticle data were only analyzed in terms of the particle sampling rate. A comparison of the single-pulse to dualpulse hit rate show ed an increase by a factor of 2.6 (250% increase) with the dualpulse configuration. This was based on the raw numbers of hits recorded using a thre shold algorithm as previously described [Carranza 2003]. In addition, all recorded si ngle-shot spectra were then post-processed using both calcium emission lines (393.4 and 396.9 nm) in an attempt to reject spectra corresponding to false partic le hits, as previously described [Hahn and Lunden 2000]. From the average spectra of all the hits, the average ratio of the peak-to-base for the 393.4 Ca II line to the 396.9 Ca II line was dete rmined. All the hits that were inside the range of the average ratio a factor of 2 were decided as acceptable hits and all those out of this range were rejected. Following this analysis, the ratio of particle hit rates with dual-pulse LIBS as compared to the single-pul se (Laser 2 only) c onfiguration was again equal to 2.6, in exact agreemen t with the previous result. Ov erall, the borosilicate glass particle experiments were consistent with the above fine-particulate experiments, verifying that the dual-pul se LIBS configuration does produce an enhanced analyte response for the micron-sized particles as well.

PAGE 68

56 3.2 Particle Lens Study An important topic when analyzing aeroso ls using LIBS is the introduction of the analyte into the sample volume. Like in the first study, often a uniformly concentrated aerosol and gas mixture flows through a large cross-sectioned tube in which the plasma volume is situated in the center. However, if the cross-section were reduced in size, a more concentrated flow could be achieved, and one would expect hit rates along with analyte response to increase. This study focu ses on the above comments, in which aerosol response and hit rates were compared for sample injection using a flow lens to reduce the flow cross-section down to 0.75 mm (.0295) diam eter versus a diameter of 0.18 of a standard tube. 3.2.1 First Attempts As mentioned in Chapter 2, experiment s were first performed by drawing the analyte through the two in jection tubes at 1 lpm without a sheath flow, the following data Table 3-1. Data representing the analyte response (Ca II) and the hit rates for the two aerosol injection methods. Steel Tube Average P/B 1000 Shot Avg 1000 Shot Stdev Hit Average Hit Stdev P/B 1 @ 393.4 nm 67.398 13.395 435.404 47.07 P/B 2 @ 396.9 nm 44.531 9.377 316.09 43.202 Hit # Average/1000 shots Hit # Stdev/1000 shots 78.929 13.714 Particle Lens Average P/B 1000 Shot Avg 1000 Shot Stdev Hit Average Hit Stdev P/B 1 @ 393.4 nm 45.353 11.305 414.494 48.748 P/B 2 @ 396.9 nm 29.753 7.275 297.622 38.058 Hit # Average/1000 shots Hit # Stdev/1000 shots 61.5 14.416

PAGE 69

57 presented in this section describes these first attempts. Data, including hit rates and analyte response for the 2m borosilicate microspheres were gathered and compared for the aerosol injection through the particle lens and the thin walled tube. The first ionized calcium (Ca II) atomic emission lines at 393.4 and 396.9 nm (0,414 cm-1 and 0,192 cm-1, respectively) were used for all sp ectra analysis. Analyte response was determined using the calculated peak-to-base of a 1000 shot averaged spectra and from the averaged spectra of all the hits acquired in the 1000 shot set. Table 3-1 shows all the averaged peak-to-base measurements and the average hit rates for both the particle lens and tubing. The number of hits as well as the analyte response were slightly better when using the tube rather than the pa rticle lens. The average hit number per 1000 laser shots for the tube was 78.9 (17.4% RSD) versus 61.5 (23.4% RSD) for the particle lens. The 1000 shot average p eak-to-base for the 393.4 Ca (II) line was 67.4 (19.8% RSD) and 45.4 (24.9% RSD) for the tu be and particle lens, respectively. Also following a similar trend, the averaged P/B for each individual hit was 435.4 (10.8% RSD) versus 414.5 (11.8% RSD) for the tu be and lens, respectively. The difference, however, for all the mentioned results were not substantial and were rarely larger than that of one standard deviation, therefore it was concluded that there was not any advantage with either injection method. Because such similar results were obtained, it was proposed that the particles were entering the sample chamber and not exiting at the same rate, instead building up and recirculating back into the plasma volume. This was verified using a green diode laser to perform light scattering to visu alize the aerosol flow. Figure 3-15 [left], shows a straight green line running across the sample chamber and in front of the particle lens. This line

PAGE 70

58 was created by light scattering from the part icles that were building up in the sample chamber. This clearly shows that particles fi lled the entire chamber, making the lens effects negligible. As mentioned in Chapter 2, to remedy this, an annular sheath flow was situated around the particle lens and tube to separate the part icles exiting the tubes with those re-circulating throughout the sample ch amber. This can also be seen in Figure 3-15 [right]. The small point of light situated in the center of the tubes is light being scattered by the particles exiti ng the particle lens, which are effectively isolated from those that are re-circulating. Light scattering particles exiting lens outlet Figure 3-15. Images depicting the calcium aerosol light scattering of a green diode laser. Looking down into the sample chamber, the beam path runs directly across the center of the pa rticle beam which is aligned in the center of the annular tube. The green line represen ts the presence of particles. The two images show the difference in part icle location without [left] and with [right] the annular sheath flow. As you can see, marked by the small spot of scattered light at the center of the tubes, th e image with the annular sheath flow [right] effectively separate d the particle exiting the lens with those that were re-circulating in the chamber. 3.2.2 Sheath Flow Experimental Results Like in the experiments w ithout the sheath flow, partic le hit rates, the peak-tobase from the 1000 shot aver aged spectra, and the aver age peak-to-base from each individual hit of the 2m borosilicate microspheres were used to quantify advantages or

PAGE 71

59 disadvantages in using the particle lens. Th e same calcium lines, Ca II lines at 393.4 and 396.9 nm (0,414 cm-1 and 0,192 cm-1, respectively), were used to quantify the hit rate and analyte response. It was believed that data from these experiments was more conclusive since the effect of the re-circulating part icles was removed. A threshold conditional analysis appro ach was applied to obtain hit counts and individual spectra when a hit occurred. This resulted in a number of hits for both the tube and the particle lens. To further eliminate any possible false hits, an extra filtering process was used. From the average spectra of all the hits it was calculated that the average ratio of the peak-to-base for th e 393.4 Ca II line to the 396.9 Ca II line was nominally 1.2 for both the tube and the particle lens. An acceptable range for this ratio was determined to be 1.2 a factor of 2 (0.6-2.4). This process reduced the hit count from 426 to 366 for the particle lens (a 14.1% reduction) and from 1769 to 1549 (a 12.4% reduction). The similar reductions proved that false hits were just as likely to occur whether a particle lens was us ed or not in s LIBS system. Table 3-2 displays hit rates, averaged P/B of each spectra corresponding to a hit, and the P/B of the averaged 1000 shot spectr a for the 393.4 nm Ca II spectral line. The data indicates no advantage when using a smalle r particle injection orifice. Hits occurred Table 3-2. Spectra and hit data comparing the tube to the particle lens. 393.4 nm Ca II Line Data Hit Rate Hit Average P/B Hit Stdev P/B 1000 Shot Average P/B 1000 Shot Stdev P/B 1/4" Steel Tube 0.0645 248.12 198.79 37.72 8.87 Particle Lens 0.0159 263.59 218.21 9.713 4.27 1.6 % of the time for the particle lens vers us 6.4 % of the time for the tube, a 75% reduction. The 393.4 Ca II line peak-to-base of the 1000 shot average spectra was 9.7 for the particle lens versus 37.7 for the tube, also about a 75% reduction. This is expected

PAGE 72

60 given the difference in hit rates between the two injection methods. Despite the decreased hit rates and 1000 shot average spectra P/B, it was still hopeful that the spectra from a hit would provide a better analyte response for th e particle lens due to the fact that the particle lens was more precise at delivering a particle into the core of the plasma, where higher temperatures and electron densities ex ist. However, this was not the case, the 393.4 Ca II line average peak-to-base ratio for every individual hit were only slightly better for the particle lens (263.6 with Stdev of 218.2) than with the tube (248 with a Stdev of 198.8), corresponding to only a 6% incr ease. With relative standard deviations nearing 80%, the slight increase in the averag ed hits P/B for the particle lens was not 0 50 100 150 200 250 300 350 400 010020030040050060070080090010001100 1/4" Tube Particle LensCountPeak-to-Base Range Figure 3-16. Histogram categorizing the peak-t o-base ratio of each individual hit for the tube and the particle lens. The overall intensity is down for the particle lens due to the smaller hit rates, but the distribution for both methods follow very similar trends.

PAGE 73

61 sufficient enough to rule that implementing th e particle lens in combination with a conditional analysis approach could produce better limits of detection. To further investigate the quality of each individual hit fo r the particle lens and tube, a histogram and probability plot were constructed and can be see in Figures 3-16 and 3-17, respectively. The relative shape and distri bution of the histograms for the tube and particle lens are very similar. This also negates the expectation that the particle lens would provide a better quality hit ra tio. If that were the case, one would expect to see the distribution weighted more for the higher peak-t o-base ranges. The probability plot seen -200 0 200 400 600 800 1000 1200 .01.11510203050708090959999.999.99 no lens lensP/B of 393.4 nm Ca II linePercent Figure 3-17. Probability plot comparing the distribution of the P/B ratios for all the individual hits collected after the conditional analysis approaches were performed.

PAGE 74

62 in Figure 3-17 also clearly demonstrates this fact. Both the particle lens and tube follow the exact same distribution up to the 50% mark of a P/B e qualing 200. After this, the particle lens begins to slightly diverge, increasing a little fast er than the lens, elucidating the slightly higher average P/B for the lens shown in Table 3-2. However, the divergence of the lens is not enough to make the case that it can contribute to better analyte response. 3.2.3 Summary and Particle Counting Results In review, the particle lens produced much lower hit rates which correlated to the much lower peak-to-base ratios of the averaged 1000 shot spectra. The average hits spectra for both injection methods produced P/ B ratios of very similar values. Each hit was analyzed in anticipation that a higher per centage of the hits would contain larger P/B ratios for the lens than the tube. This wa s not the case, which is evident in Figures 313 and 3-14. Basically, when a h it occurred for either the lens or the tubing (hits being 4 times more likely for the tube ) similar analyte response was found. To clarify the above data, particle counting experiment s were used to determine whether the same amount of particles were ex iting both the lens and the tube. Pulling 1.0 lpm of the analyte through the lens and th e tube, particles were counted using the light scattering device. Considering all pa rticles with diameters greater than 2 m, the lens produced an average particle ra te through the sample chamber of 154.16 particles/min with a standard deviation of 31.9 particles/mi n while the tube produced an average rate of 229.15 partic les/min with a standard devi ation of 57.3 particles/min. This corresponded to the particle lens tran smitting about 67% of the particles of the tube. This value of 67% transmission for the particle lens was almost identical with preliminary model predictions done by Aerodyne Research, Inc. for 3 m particles

PAGE 75

63 flowing through the lens at 1 lpm [Wormhoudt 2006]. The transmission difference is a large contributor for the discrepancy in the hit rates between the two injection methods presented in Table 3-2, and coul d be attributed to particles sticking on the inner surface of the particle lens when cross-sections we re reduced by the smaller concentric tube. Since 67% of entering particles were being transmitted through the focusing lens, the question was, what else was contributi ng to the almost 75% reduction in hit rates measured with the lens? Using a probability analysis and a Poisson distribution, while assuming the flows had uniform particle di stributions, predicted sampling rates were determined for both the lens and the tube. Eq. 3.1 represents the Poison probability (P0) that zero particles will be sampled. Subtrac ting that from one, as in Eq. 3.2, gives the probability (Pn) that any number of particles will be sampled in the plasma, where represents the average number of particles pe r plasma volume and can be calculated, as in Eq. 3.3, by multiplying the particle number density (N) by the effective plasma P0=exp(-) (3.1) Pn=1-exp(-) (3.2) =N*V (3.3) volume (V).The above transmission values along with the known flow rate of 1 lpm were used to determine the particle number density that was exiting both the lens and the tube. For the lens Nlens=0.15 particles/cc a nd for the tube Ntube=0.23 particles/cc. The calculated number densities are not representa tive of the actual number densities due to a number of reasons: (1) there was also a sheat h flow diluting the actual number densities that was not accounted for, (2) the particle counter had uncertainty, and (3) not all particles exiting the lens a nd tube were going into the c ounter (the majority actually

PAGE 76

64 entering the vacuum pump). However, when the two numbers are compared in a ratio, it is significant and can be related to the ratio of the measured sampling rates in Table 3-2. The effective statistical plasma diameter wa s previously determined as 1.3 mm [Carranza and Hahn 2002b] which was used to calculate the effective plasma volume for the tube, Vtube=0.00123 cc. For the lens, the diameter of the flow exiting was smaller than the effective plasma diameter of 1.3 mm, and ther efore the effective plasma volume had to be calculated as a cylindrical section of the plasma with a diamet er equaling 0.75 mm (diameter of the lens orifice) and a lengt h of 1.3 mm (diameter of the plasma). The calculated effective plasma volume for the lens was Vlens=5.74E-4 cc. Figure 3-18 illustrates the reduction in the effective plas ma volume when using the particle lens. Plugging in the effective plasma volumes a nd number densities into Eqn. 3.3 to find Plasma 1.3 mm 0.75 mm Thin Walled Tube Particle Focusing Lens Plasma A B Figure 3-18. The thin walled tube [A ] allowed for the maximum effective plasma volume, represented by the shaded region, to excite the aerosol samples, while the particle lens [B] reduced the effective plasma volume which aided in reducing the particle hit rates. and then into Eqn. 3.2, predicted sampling rates for each injection method were calculated as Pnlens=0.89 particles/1000 shots and Pntube=2.8 particles/1000 shots for the lens and tube, respectiv ely. This gave a ratio (Pnlens/Pntube) of about 0.31, which closely compares to the measured ratio of 0.25. This means that the combination of the

PAGE 77

65 reduced particle transmission with the reducti on in the effective plasma sampling volume, almost completely accounted for the large differences in particle sampling rates. Now that the hit rate data has been e xplained, the question now becomes, why is not the analyte response for the individual hits much larger fo r the lens than the tube? One reason could be explained through prel iminary research done by Aerodyne in which, measured particle beam profile s indicated that the particles had tendencies to disperse in an annular direction [Wormhoudt 2006] therefore creating high concentrations of particles possibly missing the hotter and more ionic core of the laser induced plasma, rendering analyte responses similar to that of the tube. Also, although the particle streamline was set to go thr ough the center of the plasma, pa rticles could be sampled at any point along the streamline path within the 1.3 mm plasma diameter. This meant that the probability of sampling a particle in the center of the plasma along the streamline is less than sampling a particle at the outer edge (before and after the center) of the plasma along the streamline. More simply stated, even though the lens points the particles in the direction of the center of the plasma, it is still unlikely to have a particle located directly in the center when the plasma is formed.

PAGE 78

CHAPTER 4 CONCLUSIONS AND PROPOSED FUTURE WORK In this study two methods were investigated to assess their pote ntial to increase the analyte response and sampling rates for laser-induced breakdown spectroscopy, notably for aerosol analysis. The two methods investigated were (1) applying a dual-pulse laser configuration system, which has been proven to increase analyte re sponse on liquid and solid samples, for gaseous and aerosol analyte systems; and (2) using a particle lens to focus aerosol particles directly into the laser-indu ced plasma in hopes that sampling rates and analyte response would be enhanced. In summary, the two studies have accomplished the following: Dual-Pulse Study Based on the current analysis, clearly the dua l-pulse LIBS approach is applicable to aerosol systems, namely the analysis of particulate-phase analytes. Under such conditions, a 4-fold analyte signal (i.e., P/B and SNR) enhancement was achieved. It is concluded that the system shows promis e as a way to improve detection limits for real-time aerosol sensing applications, which might justify the added system complexity of dual-pulse configurations for critical sensing needs. On the other hand, a rather poor signal re sponse (both P/B and SNR) was realized when applying dual-pulse LIBS for analysis of strictly gas-phase species. This result, however, along with the transmission data, provides additional insight into the physics of the plasma-analyte interac tions, further supporting the concept of preferential analyte depletion within the expanding plasma for pure gas-phase analysis that has been explor ed in earlier works. Such an effect is the opposite of solid-phase analytes, which are preferenti ally enriched, allowing for a dual-pulse scheme to enhance the analyte signal (i.e., solid phase) to noise (i.e., gaseous phase) ratio. Particle Lens Study The particle lens study rev ealed no advantage in using a more complex aerodynamic lens tube over a traditional thin walle d tube. A conditional analysis approach was 66

PAGE 79

67 used to determine sampling rates, which were reduced with the particle lens by almost 75%, subsequently reducing analyte res ponse from a 1000 laser pulse averaged spectra by almost 75%. In addition, each individual hit from the particle lens and the tube were examined and compared, revealing similar responses for both particle injection methods. Therefore, the particles that were actually sampled by the plasma produced similar signals. Particle transmission along with the re duction in the effective plasma sampling volume, due to the shrinking of the flow st ream, have been considered as the major contributors for the disparity in sampling rate s. The particle lens transmitted about 2/3 of the amount of particles transmitted by the tube alone, and reduced the effective plasma volume by nearly 50%. Reasons for the similar analyte response of the individual hits include possi ble flow alterations when exiting the lens orifice causing the particle to miss the plasma core, and the fact that sampling a particle in the center of the plasma even with a lens in place is still a low probability event. To further understand the physics behind th e data acquired, additional work must be carried out for both studies. Proposed experiments and examination include: Dual-Pulse Study Running similar procedures as in this study, but with partic le analytes with varying characteristics (i.e., size, melting point, etc.) to more completely understand the plasma-particle interactions that occur a nd to determine the circumstances when a dual-pulse system could be optimized to achieve the best detection limits. Further experiments to elucidate the preferen tial accumulation of solid particles inside the laser-induced plasma should be carried out. This could include plasma imaging with multiple cameras so that a series of images could determine the particles effect with the expanding plasma, as well as performing light scattering experiments to more clearly understand the physics behind th e achieved signal enhancement for the aerosol based analyte. Particle Lens Study Flow simulation could be run to determ ine the reason for the reduced particle transmission through the partic le lens. Also, a detailed fl ow simulation could be done at the exit orifice of the particle focusing le ns to determine if there are secondary flow effects causing the particles to disperse radially rather than in a streamline. Running experiments and calculati ons to optimize the particle lens design so that the cross-section of the particle flow exi ting the lens will match with the maximum plasma volume.

PAGE 80

APPENDIX A COMPONENTS OF SPECTROSCOPIC SYSTEMS Table A-1. List of components used in the dual-pulse study. Device Manufacturer Description Equipment Nd:YAG laser Big Sky Laser Technologies 1064 nm, 10 ns pulse, 5 Hz, 290 mJ/pulse Nd:YAG laser Continuum 1064 nm, 5 Hz, 100 mJ/pulse Spectrometer Acton Research Corporation 0.275 m spectrometer, 2400 grooves/mm, 195 nm-2800 nm iCCD Princeton Instruments Intensified CCD, 200 row chip iCCD Chiller Refrigerator re-circulator Software Custom Metal Emissions Labview Program Delay Pulse Generator Stanford Research Systems, Inc. 4 channel digital delay/pulse generator Fiber Optic Acton Research Corporation 6, high optical grade, 17 fiber bundle, 1.5 mm diameter Optics Laser 1 Optical Telescope CVI Laser Corporation x2.5, two 1064 UV grade AR lenses Laser 2 Optical Telescope CVI Laser Corporation x1.7, two 1064 UV grade AR lenses Elliptical Pierced Mirror CVI Laser Corporation UV-grade AR enhanced Laser 1 Focusing Lens CVI Laser Co rporation 100-mm focal length, 50-mm dia, UV-grade, 1064 nm AR Laser 2 Focusing Lens CVI Laser Cor poration 75-mm focal length, 50-mm dia, UV-grade, 1064 nm AR 1064 nm dichroic Mirror CVI Laser Co rporation 45 degrees, 1064 dichroic mirror, 2 diameter Aperture Newport 1 aperture Quartz Window Huntington UV-grade quartz window Collection Lens CVI Laser Corporat ion 75-mm focal length, 50-mm dia, UV-grade, 1064 nm AR 68

PAGE 81

69 Table A-2. List of components us ed in the particle lens study. Device Manufacturer Description Equipment 1064 nm frequency doubled/tripled Nd:YAG Laser Continuum Surelite II 532/355 nm, 5 Hz, 60 mJ/pulse Spectrometer Acton Research Corporation 0.275 m spectrometer, 2400 grooves/mm, 195 nm-2800 nm iCCD Princeton Instruments Intensified CCD, 200 row chip iCCD Chiller Refrigerator recirculator Software Custom Metal Emissions Labview Program Vacuum Pump Thomas Compressors and Vacuum Pumps Particle Counter Particle Counting Systems Light scattering particle counter, 0.2 lpm Particle Lens Aerodyne resear ch, Inc 60 cm long, 0.75 mm exit orifice Fiber Optic Acton Research Corporation 6, high optical grade, 17 fiber bundle, 1.5 mm diameter Optics 355 nm dichroic mirror CVI Laser Corporation 45 degree, 355 nm dichroic mirror, 2 diameter Aperture Newport 1 aperture Focusing Lens CVI Laser Cor poration Plano-convex, 100 mm UVgrade, 2 diameter Collection Lens CVI Laser Cor poration Plano-convex, 100 mm UVgrade, 2 diameter Square Pierced Mirror Rolyn Optics 100 mmx100 mmx1mm center pierced-0.5.2 diameter Quartz Window Huntington UV-grade quartz window

PAGE 82

APPENDIX B PARTICLE LENS CALCULATIONS Table B-1. Results of calculations used to determine predicted sampling rates for the particle lens and steel tube. Tube Diameter Tube Area Volume Flow Rate Particle Transmission Particle Velocity (m) (m^2) (m^3/s) (particles/s) (m/s) Particle Lens 7.50E-04 4.42E-07 1.67E-05 2.569 37.73 1/4" Tube 4.57E-03 1.64E-05 1.67E-05 3.7616 1.0152 Particle Density (N) Effective Plasma Vol. Pn (calculated hit rate) Measured Sampling rate (particles/cc) (cc) (particles/plasma volume) (particles/1000 shots) (particles/1000 shots) Particle Lens 0.1546 5.74E-04 8.88E-05 8.88E-02 1.59E-02 1/4" Tube 0.229 0.00123 2.82E-04 2.82E-01 0.064542 Ratio 31.49% 24.66% 70

PAGE 83

LIST OF REFERENCES Angel S.M., Stratis D.N., Eland K.L., Lai T., Berg M.A., Gold D.M., Fresenius Journal Analytical Chemistry 369 (2001) 320. Bilajic D., Corsi M., Cristoforetti G., Legnaio li S., Palleschi V., Salvetti A., and Tognoni E., Spectrochimica Acta Part B, (2002) 339. Buckley, S.G.; Johnsen, H.A.; Hencken, K. R.; Hahn, D.W., Waste Manage. 20 (2000) 455. Carranza J.E., Fisher B.T., Yoder G.D., and Hahn D.W., Spectrochimica Acta Part B, 56 (2001) 851. Carranza J.E., Hahn D.W., Spectrochim ica Acta Part B, 57 (2002a) 779. Carranza J.E., Hahn D.W., J. Anal. Atomic Spectroscopy 17 (2002b) 1534. Carranza J.E., Hahn D.W., Analytical Chemistry 74 (2002c) 5450. Carranza J.E., Iida K. and Hahn D. W., Applied Optics, 42 (2003) 6022. Cheng M-. D., Fuel Processing Technology, 65 (2000) 219. Ciucci A., Corsi M., Palleschi V., Rastel li S., Salvetti A., and Tognoni E., Applied Spectroscopy 53 (1999) 960. Coloa F., Lazic V., Fantoni R., Pershin S ., Spectrochimica Acta Part B: Atomic Spectroscopy 57 (2002) 1167. Corsi M., Cristoforetti G., Giuffrida M., Hildago M., Legnaioli S., Palleschi V., Salvetti A., Tognoni E., Vallebona C., Spectrochimica Acta Part B: Atomic Spectroscopy 59 (2004) 723. Cremers D. A. and Radziemski L. J., Applied Spectroscopy, 39 (1985) 57. Dudragne L., Adam Ph., Amouroux J., Applied Spectroscopy, 52 (1998) 1321. Essien M., Radziemski L.J., and Sneddon J., Journal of Analytical Atomic Spectrometry, 3 (1988) 985. Ferioli F., Puzinauskas P.V., Buckley S.G., Applied Spectroscopy 57 (2003) 1183. 71

PAGE 84

72 Fisher B.T., Johnsen H.A., Buckley S.G., and Hahn D.W., Applied Spectroscopy, 55 (2001) 1312. Gautier C., Fichet P., Menut D., Lacour J. -L., LHermite D., Dube ssy J., Spectrochimica Acta Part B: Atomic Spectroscopy 60 (2005) 265. Gleason R.L. and Hahn D.W., Spectrochimica Acta Part B 56 (2001) 419. Gornushkin I.B., Kazakov A.Y., Omenetto N., Smith B.W.,Winefordner J.D., Spectrochimica Acta Part B 59 (2004) 401. H. Kaiser, Spectrochimica Acta, 33B (1978) 551. Hahn D. W., Flower W. L., and Hencken K. R., Applied Spectroscopy, 51 (1997) 1836 1844. Hahn D.W., Applied Physics Letters, 72 (1998) 2960. Hahn D.W. and Lunden M.M., Aeroso l Science and Technology, 33 (2000) 30. Hahn D.W., Carranza J.E., Arsenault G.R., J ohnsen H.A., and Hencken K.R., Review of Scientific Instruments, 72 (2001) 3706. Hohreiter V., Ball A., Hahn D.W., J. Analytical Atomic Spectroscopy, 19 (2004a) 1289 1294. Hohreiter V., Carranza J. E ., Hahn D. W., Spectrochimica Acta Part B: Atomic Spectroscopy 59 (2004b) 327. Hohreiter V. and Hahn D. W., Anal ytical Chemistry, 77 (2005a) 1118. Hohreiter V., Hahn D.W., Spectrochimica Acta Part B: Atomic Spectroscopy 60 (2005b) 968. Hohreiter V. and Hahn D. W., Analytical Chemistry, 78 (2006) 1509. Ingle J. D. J., and Crouch S. R. Spectrochemical Analysis (1988) Prentice Hall, Englewood Cliffs, NJ. Ismail M. A., Imam H., Elhassan A., Youniss W. T., and Harith M. A., J. Analytical Atomic Spectroscopy, 19 (2004) 489. Kaiser H., Spectrochimica Acta, 33B (1978) 551. Lee W-. B., Wu J., Lee Y-. I., and Sneddon J ., Applied Spectroscopy Reviews, 39 (2004) 27. Lencioni D.E., Applied Physics Letters, 23(1) (1972) 12.

PAGE 85

73 Lindner H., Koch J., Niemax K., Analytical Chemistry, 77 (2005) 7528. Martin M. Z. and Cheng M-D., Applied Spectroscopy, 54 (2000) 1279. Neuhauser R. E., Panne U., Niessner R., Pe trucci G. A., Cavalli P., Omenetto N., Analytica Chimica Acta. 346 (1997) 37. Neuhauser R.E., Panne U., Niessner R., and W ilbring P., Fresenius Journal of Analytical Chemistry, 364 (1999) 720. Nunez M.H., Cavalli P., Petrucci G., and Omenetto N., Applied Spectroscopy. 54 (2000) 1805. Peng L. W., Flower W. L., Hencken K. R., John sen H. A., Renzi R. F., and French N. B., Process and Control Quality. 7 (1995) 39. Radziemski L.J. and Loree T.R., Plasma Chemistry and Plasma Processing, 1 (1981) 281. Radziemski L.J., Loree T.R., Cremers D.A ., and Hoffman N.M., Analytical Chemistry, 55 (1983) 1246. Radziemski L. J., Microchemical Journal, 50 (1994) 218. Radziemski L.J, Spectrochimi ca Acta Part B, 57 (2002) 1109. Sattman R., Sturm V., Noll R., Laser, J. Phys., D. Appl. Phys. 28 (1995) 2181. Scaffidi J., Pearman W., Lawrence M., Cart er J., Colston B.W., Angel S.M., Applied Optics, 43 (27) (2004) 5243. Schechter I., Analytical Scie nce and Technology, 8 (1995) 779. Schechter I., Reviews in Anal ytical Chemistry, 16 (1997) 173. Singh J.P., Yueh F.Y., Zhang H.S., and Cook R.L., Process and Control Quality, 10 (1997) 247. Smith D.C., Journal of A pplied Physics, 48 (1997) 2217. Smith B.W., Hahn D.W., Gibb E., Gornushki n I., and Winefordner J.D., KONA Powder and Particle N0 19 (2001) 25. Sneddon J. and Lee Y-I, Anal ytical Letters, 32 (1999) 2143. St-Onge L., Sabsabi M., Cielo P., Spectrochim ica Acta Part B: Atomic Spectroscopy, 53 (1998) 407. Stratis D.N., Eland K.L., Angel S.M ., Applied Spectroscopy 54 (9) (2000) 1270.

PAGE 86

74 Stratis D.N., Eland K.L., Angel S.M., Applied Spectroscopy 55 (2001) 1297. Tognoni E., Palleschi V., Corsi M., Cristofo retti G., Spectrochimica Acta Part B, 57 (2002) 1115. Tran M., Smith B.W., Hahn D.W., Wine fordner J.D., Applied Spectroscopy, 55, (2001)1455. U. S. Enviromental Protection Agency, Federal Registration 61, 77, (1996) 17357. Uebbing J., Brust J., Sdorra W., Leis F., Ni emax K., Applied Spectroscopy, 45 (9) (1991) 1419. Weyl G.M. Physics of Laser-Induced Breakdown: An Update, in Laser-Induced Plasmas and Applications edited by L.J. Radziemski a nd D.A. Cremers (1989), Marcel Dekker, New York, Chapter 1. Winefordner J. D., Gornushkin I. B., Correll T., Gibb E., Smith B. W., and Omenetto N., Journal of Analytical Atomic Spectrometry, (2004), 19, 1061. Wormhoudt J. C., Particle Lens Guide. Boston, MA: Aerodyne Research, Inc. (2006) Zhang H., Yueh F.Y., and Singh J.P., Applied Optics, 38 (1999) 14591466.

PAGE 87

BIOGRAPHICAL SKETCH Bret Windom earned his underg raduate degree in Mechanical Engineering at the University of Florida in 2004. The work presen ted here is a culmination of his Masters studies, while he is currently pursuing his doctoral degree under Dr. David Hahn at the University of Florida. 75


Permanent Link: http://ufdc.ufl.edu/UFE0016340/00001

Material Information

Title: Implementation of Aerodynamic Focusing and Dual-Pulse Configuration to Improve Laser-Induced Breakdown Spectroscopy Aerosol Particle Sampling Rates and Analyte Response
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0016340:00001

Permanent Link: http://ufdc.ufl.edu/UFE0016340/00001

Material Information

Title: Implementation of Aerodynamic Focusing and Dual-Pulse Configuration to Improve Laser-Induced Breakdown Spectroscopy Aerosol Particle Sampling Rates and Analyte Response
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0016340:00001


This item has the following downloads:


Table of Contents
    Title Page
        Page i
        Page ii
    Acknowledgement
        Page iii
    Table of Contents
        Page iv
        Page v
    List of Tables
        Page vi
    List of Figures
        Page vii
        Page viii
        Page ix
        Page x
    Abstract
        Page xi
        Page xii
    Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
    Experimental methods
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
    Results and discussion
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
    Conclusions and proposed future work
        Page 66
        Page 67
    Appendices
        Page 68
        Page 69
        Page 70
    References
        Page 71
        Page 72
        Page 73
        Page 74
    Biographical sketch
        Page 75
Full Text












IMPLEMENTATION OF AERODYNAMIC FOCUSING AND A DUAL-PULSE
CONFIGURATION TO IMPROVE LASER-INDUCED BREAKDOWN
SPECTROSCOPY AEROSOL PARTICLE SAMPLING RATES AND ANALYTE
RESPONSE















By

BRET C. WINDOM


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2006





























Copyright 2006

By

Bret C. Windom















ACKNOWLEDGMENTS

First and foremost, I would like to thank Dr. Hahn for his guidance and leadership

provided during the current study. Because of his enthusiasm and expertise in the subject

I have learned alot in the short time I have been here. Secondly, I would like to thank my

lab mates for providing assistance in my research and course work, and also, for creating

an enjoyable and stimulating work atmosphere.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ......... .................................................................................... iii

LIST OF TABLES ............. ..... ......................... .......... ............ vi

LIST OF FIGURE S ......... ..................................... ........... vii

ABSTRACT .............. ......................................... xi

CHAPTER

1 IN TR OD U CTION .................. ............................. ....... ...... .............. .

1.1 O overview ...................................... .....................................1
1.2 Laser Induced Breakdown Spectroscopy (LIBS) Process............... ...............2
1.2 .1 P lasm a F orm ation ...................................................... .......................... .. .2
1.2.2 Temporal Evolution and Decay of a Laser-Induced Plasma........................3
1.2.3 A tom ic Spectral Em mission Collection ..........................................................5
1 .3 L IB S S ig n al ....................................................... ................ .. 5
1.3.1 A nalyte R response ............................ .................. .............. ............ ..
1.3.2 Calibration .................................... .......................... ... .........6
1.3.3 C alibration Curve ................... .. ...................... .. ....... .......... ...... .
1.3.4 Lim its of D etection.......... ...................................... ...... .......... ........
1.4 A erosol Based LIB S ...................... .... ................ ..... .............. .. .. ...10
1.4.1 M ass Concentration and Size Detection................. ............ ... ..............10
1.4.2 Continuous Em mission LIB S .................................. ............ ................... 11
1.4.3 Conditional A analysis ........................................................................... 12
1.4.4 Fundam mental A erosol Studies................................. ......... ..............13
1.5 Dual-Pulse LIBS ............................. ... .......... ...... .. ................. 15
1.6 O bjective................................................. 17

2 EXPERIM EN TAL M ETHOD S.......................................................... ............... 19

2.1 LIBS Experimental Setup and Procedure.........................................................19
2.1.1 Plasm a Creation M ethods................................................. ...... ......... 19
2.1.2 E m mission C ollection......................................................... ............... 20
2.1.3 Aerosol Generation................................................. 21
2.2 Dual-Pulse LIBS: Gas and Aerosol Study..........................................................24
2.2.1 Laser Configuration........................................................ ............. 24










2.2.2 Experim ental Analyte Generation ................................... .................26
2.2.3 Spectral M easurem ents......................................... .......................... 27
2.2.4 Transm mission M easurem ents ........................................... ............... 29
2 .3 P article L ens Study .............................. .......... ... ...........................................29
2.3.1 Laser Configuration.......... ..... ........... .................. 29
2.4.2 A nalyte G generation .......................................................... ............... 30
2.4.3 Spectral M easurem ents........................................ ........................... 32
2.4.4 Particle Counting M easurements...................................... ......... .......... 32

3 RESULTS AND DISCUSSION .......... ..... ........... .................. 33

3.1 D ual-Pulse LIB S Study........... ................. ........... ................ ............... 33
3.1.1 Transm mission Experim ents ........................................ ...... ............... 33
3.1.2 Spectral Analysis of Gaseous Analyte ..................... ...................38
3.1.3 Spectral Analysis of Fine Calcium-Based Aerosol Analyte....................47
3.1.4 Spectral Analysis of Borosilicate Glass Microspheres............................53
3.2 P article L ens Study ............................... .... .......... .... .... ...... ...... 56
3 .2 .1 F first A ttem pts ................................................... ....... ... ......... .. .. 56
3.2.2 Sheath Flow Experimental Results .......... .............. ........................58
3.2.3 Summary and Particle Counting Results..................... .................62

4 CONCLUSIONS AND PROPOSED FUTURE WORK......................................... 66

APPENDIX

A COMPONENTS OF SPECTROSCOPIC SYSTEMS...............................................68

B PARTICLE LENS CALCULATIONS ............................................................. 70

L IST O F R E F E R E N C E S ....................................................................... ... ................... 7 1

B IO G R A PH IC A L SK E TCH ..................................................................... ..................75




















v
















LIST OF TABLES


Table page

3-1 Data representing the analyte response (Ca II) and the hit rates for the two
aerosol injection m ethods ......... ......... ................ ...................... ............... 56

3-2 Spectra and hit data comparing the 4" tube to the particle lens. ..........................59

A-i List of components used in the dual-pulse study. ............. ..................................... 68

A-2 List of components used in the particle lens study.............................. ...............69

B-l Results of calculations used to determine predicted sampling rates for the
particle lens and /4" steel tube ................. ......... ................... 70















LIST OF FIGURES


Figure page

1-1 Spectral line representation with increasing atomic concentration (a-*h). Once
the concentration reaches a certain amount, the emission intensity reaches a limit
given by Bb m (blackbody radiator). The wings then start to widen causing
calibration curves to flatten [Ingle and Crouch 1988]................................................8

1-2 Calibration curve of growth. At first, the curve increases linearly until a certain
concentration is achieved and the maximum linear atomic emission intensity is
reached, equal to that of a blackbody radiator. At this instance self-absorption
begins, and the curve begins flatten out. .............. ..................... ...................... 9

1-3 Illustrations representing two setups used in many of the dual-pulse studies; [A]
An orthogonal configuration and [B] An in-line configuration. ...........................16

2-1 Optical setup for the delivery of the laser irradiance to the sample volume
resulting in the creation of a plasma................................................. ................ 19

2-2 Optical setup for the collection of plasma emission from the sample volume.........21

2-3. General schematic of the aerosol generation system. The co-flow rates were
those used in the dual-pulse study [Hahn 2001]. ............................................. 22

2-4 Chart displaying Cu analyte counts of remnant solutions from a nebulizer as a
function of nebulization time. Error bars represent the standard deviation. ............23

2-5 Experimental apparatus for single and dual-pulse LIBS configurations..................24

2-6 Schematic representing the trigger setup to achieve laser pulses fired
simultaneously in time or at a variable delay. Delay generator 1 set to 77.75 [ts
would result in the lasers firing simultaneously. Adjusting the variable delay of
generator 1 would result in delaying laser 2 pulse. .............................................26

2-7 The optical setup (top view) for the particle lens experiments. Like the dual-
pulse setup, the plasma emission was back collected. The aerosols were drawn
through the particle lens into the plasma by a vacuum pump situated on the
opposite side of the 5-w ay cross. ........................................ ......................... 30









2-8 Simple schematic of the particle lens. Aerosols flow through a series of radial
decreasing concentric tubes before exiting through an outlet causing a streamline
exit into the laser induced plasm a ................................................ ....... ........ 31

3-1 Laser 2 transmission as a function of dual-pulse laser delay times for the pure air
and fine calcium aerosol sample streams. The horizontal line represents the
transmission of Laser 2 alone (i.e., single pulse LIBS). Note the plot is linear for
delay tim es less than 100 ns. ............................................ ............................ 34

3-2 Spectra showing the two nitrogen atomic emission lines at 491.4 nm and 493.5
nm. The spectra correspond to dual-pulse LIBS with 500 ns delay (lower
spectrum) and single-pulse (Laser 2 only) LIBS (upper spectrum). Both spectra
have the sam e scale. ........................................ ............................ 38

3-3 The 493.5-nm nitrogen emission line peak-to-base ratio measurements for the
pure air sample as a function of dual-pulse laser delay times. The P/B ratios are
also shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam
blocked). The dashed horizontal line represents the average P/B ratio for Laser
2 only. A representative error bar is included on the 100 ns Laser 1 only mark. ....40

3-4 The 493.5-nm nitrogen emission line signal-to-noise ratio measurements for the
pure air sample as a function of dual-pulse laser delay times. The SNR are also
shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked).
The dashed horizontal line represents the average SNR ratio for Laser 2 only.......41

3-5 Spectra showing the oxygen (I) triplet at 394.7 nm. The spectra corresponds to
dual-pulse LIBS with 500 ns delay and single-pulse (Laser 2 only) LIBS. Both
spectra have the sam e scale .............................................. ............................ 42

3-6 Oxygen peak-to-base measurements for the filtered air sample as a function of
dual-pulse laser delay times. The average peak-to-base ratio for Laser 2 only is
represented by the dashed horizontal line. .................................... .................43

3-7 The 397.4-nm oxygen emission line signal-to-noise ratio measurements for the
pure air sample as a function of dual-pulse laser delay times. The S/N are also
shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam blocked).
The dashed horizontal line represents the average S/N ratio for Laser 2 only.........44

3-8 Image detailing the proposed mechanism, in which an expanding plasma volume
will force the nano-scale gas phase particles outward depleting their
concentration in the plasma core [Hohreiter and Hahn 2005a].............. ...............46

3-9 Spectra showing the Ca II atomic emission lines at 393.4 and 396.9 nm for both
the dual-pulse configuration with a 250-ns delay, and for Laser 2 only. Both
spectra have the same intensity scale, and the dual-pulse spectrum has been
shifted upward by 400 counts for clarity....................................... ............... 47









3-10 Spectra showing the Ca II atomic emission lines at 393.4 and 396.9 nm for both
the dual-pulse configuration with a 750-ns delay, and for Laser 2 only. Both
spectra have the same intensity scale, and the dual-pulse spectrum has been
shifted upward by 400 counts for clarity....................................... ............... 48

3-11 Spectra showing the Ca II atomic emission lines at 393.4 and 396.9 nm for both
the dual-pulse configuration with a 50-ts delay, and for Laser 2 only. Both
spectra have the sam e intensity scale. ........................................... ............... 49

3-12 The 393.4 -nm calcium II emission line peak-to-base ratio measurements for the
fine calcium aerosol sample as a function of dual-pulse laser delay times. The
P/B are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2 beam
blocked). The dashed horizontal line represents the average P/B ratio for Laser
2 only. 50

3-13 The 393.4 -nm calcium II emission line signal-to-noise ratio measurements for
the fine calcium aerosol sample as a function of dual-pulse laser delay times.
The SNR are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2
beam blocked). The dashed horizontal line represents the average SNR for
L aser 2 only .............. .. ....... ...................... ......................... 51

3-14 This image depicts the basis for the increase analyte response in a dual-pulse
configuration for an aerosol sample. The larger aerosol particles resist the radial
force exerted on them by the expanding plasma volume created by laser 1 and
remain inside the core while the smaller gas phase particle are expelled, thereby
creating a more aerosol concentrated volume that is awaiting the plasma from
laser 2 [Hohreiter and Hahn 2005a]. .............................. ................................. 52

3-15 Images depicting the calcium aerosol light scattering of a green diode laser.
Looking down into the sample chamber, the beam path runs directly across the
center of the particle beam which is aligned in the center of the 12" annular tube.
The green line represents the presence of particles. The two images show the
difference in particle location without [left] and with [right] the annular sheath
flow. As you can see, marked by the small spot of scattered light at the center of
the tubes, the image with the annular sheath flow [right] effectively separated
the particle exiting the lens with those that were re-circulating in the chamber.....58

3-16 Histogram categorizing the peak-to-base ratio of each individual hit for the /4"
tube and the particle lens. The overall intensity is down for the particle lens due
to the smaller hit rates, but the distribution for both methods follow very similar
trends. 60

3-17 Probability plot comparing the distribution of the P/B ratios for all the individual
hits collected after the conditional analysis approaches were performed ...............61

3-18 The thin walled 4" tube [A] allowed for the maximum effective plasma volume,
represented by the shaded region, to excite the aerosol samples, while the









particle lens [B] reduced the effective plasma volume which aided in reducing
the particle hit rates. ....................... ...................... ..................... .. .... .. 64

















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

IMPLEMENTATION OF AERODYNAMIC FOCUSING AND A DUAL-PULSE
CONFIGURATION TO IMPROVE LASER-INDUCED BREAKDOWN
SPECTROSCOPY AEROSOL PARTICLE SAMPLING RATES AND ANALYTE
RESPONSE

By

Bret C. Windom

December 2006

Chair: David Hahn
Department: Mechanical Engineering

This study focused on two alternative methods to increase the analyte response

using laser induced breakdown spectroscopy (LIBS) specifically on gaseous and aerosol

phase analytes. The first, using a dual-pulse LIBS configuration to enhance analyte

response of elements in an air sample and aerosol particles consisting of calcium; and

second, using a particle focusing lens as a way to feed aerosols into the laser-induced

plasma to produce higher particle hit rates and/or an enhanced analyte response.

Dual-pulse LIBS has previously demonstrated to significantly enhance the analyte

peak-to-base and signal-to-noise ratios for solid and liquid phase analytes. This study

focused on the effects of an orthogonal dual-pulse laser configuration on the atomic

emission response for both purely gaseous and calcium-based aerosol samples. The

gaseous sample consisted of purified (i.e., aerosol free) air, from which nitrogen and

oxygen spectral emission lines were analyzed. Measurements for the gaseous system









resulted in no notable improvements with the dual-pulse configuration as compared to

single-pulse LIBS. Experiments were also conducted in purified air seeded with calcium-

rich particles, which revealed a marked improvement in calcium atomic emission peak-

to-base (-2-fold increase) and signal-to-noise ratios (-4-fold increase) with the dual-

pulse configuration. In addition to increased analyte response, dual-pulse LIBS yielded

an enhanced single-particle sampling rate when compared to conventional LIBS.

Transmission measurements with respect to the plasma-creating laser pulse were

recorded for both single and dual-pulse methods over a range of temporal delays. In

consideration of the spectroscopic and transmission data, the plasma-analyte interactions

realized with a dual-pulse methodology are explained in terms of the interaction with the

initially expanding plasma, which differs between gaseous and particulate phase analytes.

A particle-focusing lens is a way to reduce the cross-section of the flow and eject

a narrow streamline of particles. This study focused on the notion that if a streamline of

particles could be injected directly into the center of a laser-induced plasma by a particle

lens, higher aerosol sampling rates and better analyte response could be achieved.

Following all experiments and analysis, it was determined that a particle lens with exit

cross-section diameter of 0.0295" showed a decrease in particle hit rates and an

unchanged analyte response when compared to that of a 14" diameter standard thin walled

tube. The average hit rates decreased about 75% with the particle lens and resulting

analyte response of all individual hits for the two injection methods were nearly the same.

The lower particle hit rate with the focusing lens was partially attributed to a reduction in

particle transmission (as measured with a particle counter), while other effects such as

changes in the effective plasma sampling volume were also considered.
















CHAPTER 1
INTRODUCTION

1.1 Overview

Laser-induced breakdown spectroscopy (LIBS) also referred to as laser-induced

plasma spectroscopy (LIPS) had its inception in the 1960s with the development of the

first laser. As instrumentation progressed and became less expensive, LIBS grew in

popularity as an analytical technique. Since the 1980s research and published literature

has increased, including many literature reviews [Schechter 1997, Smith 2001, Tognoni

2002, Radziemski 2002, Lee 2004, Winefordner 2004], as LIBS has begun to obtain

recognition on the likes of other analytical techniques such as inductively coupled

plasma-mass spectrometry (ICP-MS) and inductively coupled plasma-atomic emission

spectrometry (ICP-AES) [Winefordner 2004].

A LIBS system entails tightly focusing a pulsed laser beam onto the medium of

interest, in which the resulting increase in irradiance acts in disassociating molecules and

creating a volume with high ion and free electron densities along with high temperatures

(in excess of 30,000 K) called a plasma or spark. As the plasma temperatures cool and the

ionization degree falls, the electronically excited atoms within the plasma relax and emit

elemental specific radiation (i.e., atomic emission) which can be analyzed with a

spectrometer in combination with an intensified charge coupled device (ICCD) detector.

The measured spectral lines can be used to carry out measurements relating species

existence, concentration, and mass. The LIBS technique has the ability to perform









analysis on gases, liquids, solids, and, an important research topic, aerosol systems.

Because of simple and relatively inexpensive instrumentation, its aptness for real-time in

situ measurements, and its prospect to being applied to micro-systems, LIBS shows huge

promise to becoming a prominent analytical technique in the future.

1.2 Laser Induced Breakdown Spectroscopy (LIBS) Process

1.2.1 Plasma Formation

A laser-induced plasma, qualitatively characterized by a bright spark followed by

an audible shock wave, is created when the irradiance of the laser beam exceeds the

dielectric strength or the breakdown threshold of the medium [Weyl 1989]. For example,

air has a breakdown threshold that can range from 90-1600 GW/cm2 for a 1064 nm laser

source, and from 30-380 GW/cm2 for 266 nm lasers [Smith 2001]. Beam irradiance can

be increased by focusing the beam to its diffraction limit, by using short duration pulses

(- ns to fs), and by simply increasing the laser pulse energy. The two mechanisms that

primarily result in plasma formation are multi-photon ionization and cascade ionization,

also known as avalanche ionization.

Multi photon ionization is the process when electrons are excited by some

incident radiation source, to the point that their energy becomes great enough to break

away from the atom and become free electrons. Typically, for the process to occur with

visible and IR radiation souces, the energy from multiple photons are needed, therefore

the probability for occurrence is based on photon density. Multi photon ionization can be

described in the following equation.

M + n(hv) e- + M+ (1.1)

In which M is the neutral atom and n(hv) is n number of photons when combined create a

free electron and an ionized atom. The number of photons needed, n, is dependent on the









energy of a single photon. For example, since a photon with a wavelength of 532 nm is

more energetic than one of 1064 nm (2.33 eV vs. 1.17 eV), the 532 nm light would be

able to achieve ionization with less photons than that of 1064 nm light (i.e., in order to

ionize N2, n532 = 4-6 photons, while n1064 = 8-12 photons).

The second mechanism responsible for plasma creation is cascade ionization.

Since free electrons are not quantized, the radiative energy absorbed from the incident

laser by the free electrons are converted into translational energy which causes them to

collide with neutral atoms or molecules resulting in their ionization and another free

electron, represented in the following equation.

e- + M 2e + M (1.2)

The relationship above shows that the free electrons are doubled with every

occurrence of cascade ionization, therefore causing an exponential increase in the number

of free electrons. Cascade ionization is thought to be the prominent mechanism in plasma

creation, but since free electrons are needed, multi photon ionization is said to initiate the

breakdown process. However, due to the higher energy per photon, for shorter

wavelength laser sources (<1 .im), such as the 266 nm (4.67 eV)source, the process of

multi photon ionization is more important in the plasma creating process, whereas it takes

a back seat to cascade ionization when dealing with longer wavelength sources (i.e., 1064

nm) [Martin and Cheng 2000].

1.2.2 Temporal Evolution and Decay of a Laser-Induced Plasma

Although plasma characteristics can vary based on laser power, laser wavelength,

optical setup, sample material and concentration, the following represents the evolution

and life of a typical laser-induced plasma. By the end of the laser pulse, plasma

temperatures and free electron number densities rise to their maximum, as much as









40,000 K and 1018-1019/cm3, respectively, as approximately 50% to 90% of the laser

pulse energy is coupled into the plasma. It should be noted that radiant emission from the

plasma (also known as the continuum radiation) in the ultraviolet-visible range occurs

shortly after the laser pulse reaches the focal spot and is representative of emission from a

blackbody. The spectral distribution of the continuum can vary with plasma temperature,

for example, a hotter plasma will shift the continuum to smaller wavelengths and vice

versa for cooler plasmas.

Continuum emission is the result of two processes, recombination (free-bound) and

Bremsstrahlung (free-free). Recombination occurs as an ion captures a free-electron

resulting in emission of the excess energy as the free electron becomes bound.

Bremsstrahlung emission occurs as a free electron decelerates near and then past an ion

causing an energy adjustment resulting in the emission of a photon equal to the energy

change.

Generally, plasma temperatures and free electron densities peak by the tail end of

the laser pulse, although the plasma continues to grow in volumetric size and continues

its strong broadband continuum emission, which dominates atomic lines early in time.

After a few microseconds the plasma has significantly cooled due to energy transfer from

radiation, recombination, and quenching, and the continuum effects become less

dominate, therefore spectral atomic lines become detectable. Since the bound electrons

are quantized, as they start to relax to electronic ground states, photons, with frequencies

inherent to each atom present in the plasma volume are emitted (similar to laser induced

fluorescence). The optimal delay after the laser pulse in which atomic lines are best









resolved over the continuum emission is elemental dependent and can range from 1 to 30

gs (Buckley 2000).

After 30 to 50 gts, the neutral atoms start to combine, creating molecules which can

emit at their own emission bands. Emission can occur up to about one hundred

microseconds after the initial pulse, but is hard to detect after about 50 gts due to the low

signal levels. After a couple hundred microseconds, the plasma has essentially dissipated

and returned to near ambient conditions.

1.2.3 Atomic Spectral Emission Collection

Since light from the plasma emits equally in all directions, the collection devices

can be arranged through multiple configurations, but certain ones do have their

advantages. In most cases the emitted light is focused onto the tip of a fiber optic bundle

and transmitted to a spectrometer in which the light is diffracted and dispersed so that a

specific wavelength range of the emitted light is analyzed. The dispersed light is then

applied to the surface of an intensified charged coupled device (ICCD) creating a

spectrum, so that intensities of specific wavelengths representing species in the plasma

can be analyzed.

1.3 LIBS Signal

1.3.1 Analyte Response

When analyzing spectra, one can qualitatively determine the existence of certain

elements through the existence of atomic spectral lines, but this is only a fraction of what

makes LIBS a robust technique. It is known that the intensity of the atomic signal is

proportional to the number of emitting atoms/ions. This is important because it allows

LIBS to be a quantitative technique to measure analyte concentration.









So what is the best way to represent signal intensities so that concentration

measurements can be extracted? Rather than measuring the absolute intensity, which is

the combination of the background (continuum emission) and the atomic signal, a better

and more widely used measurement includes taking the integrated peak above the

continuum and normalizing it by the intensity of the continuum [Hahn 1997]. This ratio is

called the peak-to-base (P/B) and is advantageous due to the ability to minimize shot-to-

shot variations by normalization to the continuum [Carranza and Hahn 2002a]. Another

popular measurement among LIBS researchers to quantify analyte response is the signal-

to-noise ratio (S/N). In this measurement, the analyte signal, which may be the full width

integrated emission peak area, is divided by the spectral noise, which may be represented

by the root-mean-square (RMS) of the continuum intensity adjacent to each peak. The

RMS describes the average square of the deviation of the continuum from a linear fit of

itself, and is representative of the smoothness of the continuum region, which therefore is

a measure of analyte signal limitation [Ingle and Crouch 1988].

1.3.2 Calibration

In most cases when LIBS is applied to detecting amounts and concentrations of

certain species within a medium, calibration must be made, with the exception attributed

to studies such as those by Ciucci [1999] and Bilajic [2002] in which calibration free

LIBS procedures were developed. Calibration methods include analyzing spectra from

standards, which contain varying concentrations of analyte within a medium identical as

possible to that of the sample for which the concentration information is sought. By using

the same material in the standards as in the sample of interest (i.e., matching the matrix),

plasma matrix effects can be avoided. In general, matrix effects are alterations in analyte

emission due to the changes in plasma characteristics such as temperature, electron









density, quenching of atomic spectral lines by other species, and the loss of emitting

elements due to recombination. For example, significance of these effects was shown in a

study done by Gleason and Hahn [2001], in which mercury spectral lines were shown to

differ as the concentration of oxygen in the surrounding gas was altered. They concluded

that the reduction in the recorded mercury emission was due to molecular oxygen species,

mainly 02 and NO, quenching the mercury atomic signal. Also, Ismail [2004]

demonstrated such plasma matrix effects, showing that the detection limits of manganese,

silicon, magnesium, and copper varied when they were embedded in an aluminum

standard versus a steel standard.

1.3.3 Calibration Curve

Calibration curves that are linear are preferred when performing concentration

analysis using LIBS. However, nonlinearities are not uncommon, and do occur for

various reasons such as exceeding the concentration limit, quenching, and recombination.

Mentioned before, as the concentration of the analyte increases so too does the intensity

of the atomic signal. As the emission intensity of the analyte grows, it can reach an

intrinsic limit equivalent to the emission of a blackbody radiator from Plank's law. This

then causes the wings of the atomic line to widen, Figure 1-1, and can be thought in terms

of line broadening or self-absorption.

At low analyte concentrations, the signal response grows proportionally with the

number of analyte atoms. When the atomic concentration becomes too high, the emission

irradiance of a blackbody emitter is reached, and subsequently an increasing fraction of

the emitted radiation is self-absorbed by atoms in lower energy states, and the curve starts



















B,

b











Figure 1-1. Spectral line representation with increasing atomic concentration (a--h).
Once the concentration reaches a certain amount, the emission intensity
reaches a limit given by Bb m (blackbody radiator). The wings then start to
widen causing calibration curves to flatten [Ingle and Crouch 1988].

to level out, as seen in Figure 1-2 [Ingle and Crouch 1988]. Therefore, it is important to

perform LIBS measurements in the concentration range that gives a linear response,

which is typically on the order of under 10%, indicated by the concentration region to the

left of the dotted line in Figure 1-2.

Other sources that affect the calibration curve of growth include quenching and

recombination. Quenching is a non-radiative decay of atomic emission. It is species

specific and is a function of the analyte concentration as well as the concentrations of the

constituents of the plasma matrix. Recombination, like quenching, is species specific and

is also a function of analyte concentration. It occurs when the analyte and another matrix

containing species have a high affinity for each other resulting in non-radiative energy











Saturation
Regime


Signal
Response
(P/B or
S/N)








Concentration

Figure 1-2. Calibration curve of growth. At first, the curve increases linearly until a
certain concentration is achieved and the maximum linear atomic emission
intensity is reached, equal to that of a blackbody radiator. At this instance
self-absorption begins, and the curve begins flatten out.

loss due to their combination (e.g., Hg + O HgO). The last two sources of nonlinearity

may or may not be important, but should be at least considered when investigating matrix

effects.

1.3.4 Limits of Detection

When looking at a calibration curve, a figure of merit which should be explained is

the limit of detection (LOD). The LOD is the smallest concentration of analyte that can

be trusted as present in a sample with a specified level of confidence [Ingle and Crouch

1988]. It should not be confused with the sensitivity, which is the slope of the calibration

curve [Kaiser 1978]. The LOD is important in quantifying an analytical technique and is

the measurement that expresses advancements in a technique. The LOD is defined by the

following relationship:

LOD= Ko/S (1.3)









Where K is the safety factor (most commonly 2 or 3), a is the standard deviation of the

background signal (taken from a sample containing no analyte or the spectral RMS of the

noise), and S is the calculated sensitivity of the calibration curve (i.e., slope).

1.4 Aerosol Based LIBS

Though LIBS is a popular technique for applications in liquids and solids, the

gaseous phase, including aerosols, was the basis of the following study and is therefore

reviewed in detail. An aerosol is defined as solid or liquid particles suspended in a

gaseous medium. Aerosol particles range from 1 nm to 100 [im, with the particles on the

lower end of the size distribution (< 2.5 [m) receiving more recent research attention due

to their increased health implications [US EPA 1996]. Many techniques based on mass

spectrometry principles have been used to determine aerosol characteristics, but more

recently LIBS has become a candidate analytical technique in the field of individual

particle analysis. Due to its small sample size (i.e., plasma volume) in which single

particle analysis can be accomplished coupled with statistical calculations, studies have

demonstrated LIBS as an applicable species, mass concentration, and size detector in the

laboratory and in the field.

1.4.1 Mass Concentration and Size Detection

Radziemski and co-workers throughout the 1980s developed time-resolved LIBS to

overcome high continuum effects at early times and optimize their analyte signal, where

they were able to detect chlorine and phosphorous aerosols down to 60 and 15 ppm

respectively [1981]. Later, Essien [1988] detected traces of cadmium, lead, and zinc

down to detection limits of 0.019, 0.21, and 0.24 [g/g, all of which were below the

permissible limit reported by the Occupational Safety and Health Administration

(OSHA). In their study, a liquid solution with known metal concentration was nebulized









to form an air stream containing a sub-micron range of aerosols. Other metal aerosols

have been analyzed and detection limits determined, such as As, Be, Cd, Hg, Pb, Sb and

Cr [Zhang 1999, Neuhauser 1999]. Also, elements such as F, Cl, and C [Dudragne 1998]

along with compounds including phosphine (PH3), arsine (AsH3), and fluorine (CF3H)

have been investigated in similar manners [Sneddon and Lee 1999, Peng 1995, and Singh

1997].

Not only have concentration measurements been studied, but also size data has

been reported in the literature [Hahn 1998]. Hahn and Lunden [2000] calculated a size

histogram for sub-micron to micron sized calcium and magnesium aerosols. Their finding

proved that LIBS was a valid technique when sizing aerosols, as their measurements

closely corresponded to those by a light scattering technique. They determined a lower

size LIBS detection limit for the aerosols of 175 nm, and that size measurements could be

precisely made even in the presence of other aerosol types. Size measurements were

made by first determining the concentration of the aerosol by means of comparing the

unknown to a known calibration curve. Next, the absolute mass was calculated using the

determined mass concentration and the known plasma volume. Lastly, by knowing or

assuming the composition of the particle and its density, the equivalent diameter was

evaluated.

1.4.2 Continuous Emission LIBS

Because of the fast response time possible with LIBS, in situ measurements are one

of the biggest advantages in aerosol detection, and therefore explain much research

involving continuous emission studies [Zhang 1999, Neuhauser 1999, Nunez 2000,

Ferioli 2003]. Carranza [2001] studied atmospheric air, from which concentrations of

elements (aluminum, calcium, magnesium, and sodium) where measured over a 6 week









period spanning the 4th of July holiday. Using conditional analysis (methods explained

later) they found an order of magnitude increase in aluminum and magnesium during the

days adjacent and containing the holiday as a result from the increased fireworks activity.

Single shot analysis was also made to measure particle diameters, and a histogram was

created yielding sub-femtogram mass limits of detection for calcium, and around 2-3

femtograms for magnesium and sodium based particles. Buckley [2000] performed

continuous emission studies of various toxic metals on location at two waste incinerators.

Detection limits ranged from 2-100 gg/dcsm for Be, Cd, Cr, Hg, and Pb. A conditional

analysis was also used to improve detection limits over that of more traditional LIBS

procedures.

1.4.3 Conditional Analysis

Typically ensemble averaging of spectra from 100s to 1000s of laser pulses is used

to account for the random noise effects from the instrumentation [Radziemski 1994], thus

providing a decrease in the spectral noise. This is useful when the element to be analyzed

is constantly inside the plasma volume (e.g., a gas phase element). However, when

working with aerosols, especially at concentrations near the detection limit, most of the

plasmas created are absent of the analyte of interest. This is evident in the statistics

offered by Hahn [1997], in which they calculated sampling rates as low as approximately

0.1 % corresponding to about 5 particles/cm3. This means that a total ensemble average

would decrease the spectral response (peak-to-base and/or signal-to-noise) due to the fact

that most of spectra being averaged only contain continuum information and no elemental

spectral emission lines. To rectify these issues, Hahn [1997 and 2000] proposed a

conditional analysis approach that would discard the spectra that did not contain atomic

information while retaining and averaging spectra that contained spectral lines of interest.









Spectra were retained if the ratio of the atomic emission intensity and the adjacent

continuum signal exceeded a given threshold. This approach allowed for particle hit rates

to be determined as well as improving detection limits to the parts-per-trillion. Others

have incorporated similar ideas into their research including Schechter [1995], Cheng

[2000], Martin and Cheng [2000], and Carranza and Hahn [2002a].

1.4.4 Fundamental Aerosol Studies

To understand variations in spectral intensity and particle plasma interactions,

fundamental single particle analysis studies have been performed. Specifically, single

particle vaporization and their respective signal response as a function of aerosol size has

been studied. Cremers and Radziemski [1985] collected beryllium particles on filters

allowing for the collection of sizes ranging from 50 nm to 15 inm. Spectral information

was recorded on beryllium particles throughout the range of sizes. They found that the

analyte response of the 15 .im particles strayed from the linear increase that occurred for

the smaller particles as size was increased. They attributed this to the fact that once the

size reached about 10 rim, incomplete vaporization occurred affecting the analyte

response. This was the accepted upper-size limit for many years. Carranza and Hahn

[2002c] performed a series of experiments in which silica micro-spheres ranging in

diameter from 1 .im to 5 .im were analyzed and single shot spectra were recorded. A

linear increase in analyte response was detected up to diameters equaling 2.1 .im after

which the increasing diameters caused the analyte response to flatten. They also believed

that this was due to incomplete vaporization of the aerosol once diameters reached 2.1

rLm. This study also concluded that particle vaporization was more dependent on the

plasma-particle interactions rather than a laser-particle interaction, due to the fact that the

latter interaction would produce sampling rates much lower than were observed. It was









theorized that some plasma-particle interaction, such as thermophoretic forces and vapor

expulsion, was responsible for the inconsistencies found as particle diameters were

increased above the upper size limit. Hohreiter [2006] expanded on the theory established

by Carranza, in their study which included a combination of plasma imaging and

spectroscopy. Through there methods, Hohreiter determined that there is a finite amount

of time after the laser pulse when the plasma is energetic enough to disassociate particles

(about 15 ts for their study). This means that a particle with sufficient size may not be

fully disassociated, due to the fact that there is a limited amount of time for disassociation

to occur, thus agreeing with the aforementioned literature by Carranza and Hahn.

Other fundamental studies have focused on the forces produced by the expanding

plasma volume and the affect they have on the aerosol and gaseous analytes. This is

important to understand so that attempts in plasma modeling [Gornushkin 2004] can

advance. Historically, it was assumed that atomic emission was independent of the

analyte source, and that different sources would generate similar responses as long as

equal concentrations of analyte were fully disassociated [Dudragne 1998 and Essien

1988]. For example it was shown that SF6 and HF yielded identical fluorine signals as

long as the mole fractions of fluorine were the same [Tran 2001]. However, in a study by

Hohreiter and Hahn [2005a], five equally concentrated different forms of carbon analytes

were examined (2 in the solid state and 3 in the gas phase) and were found to have very

different spectral emission responses despite almost identical plasma characteristics (i.e.,

temperature and electron density). The solid state analyte had a much stronger response

than the three gas carbon analytes, which was believed to be attributed to a physical

interaction between the analyte and the expanding plasma. More specifically, the









expanding plasma was affecting the smaller lighter gas phase analyte, forcing it to the

perimeter and thereby reducing the amount of analyte present within the laser induced

plasma, and therefore decreasing the spectral response. Oppositely, the larger aerosols

withstood the force from the plasma expansion remaining engulfed in the plasma

providing strong emission lines. This physical plasma-particle phenomenon will be

discussed more in later chapters and can be used to explain some of the signal

improvements found in the current study.

1.5 Dual-Pulse LIBS

Dual-pulse LIBS has been previously investigated to a large extent on solid and

liquid phase analytes, where it has been demonstrated to significantly enhance atomic

emission signal intensity, and more importantly, to enhance the analyte peak-to-base and

signal-to-noise ratios. While the first use of an additional laser pulse to reheat the plasma

dates back to the early 1990s [Uebbing 1991], in recent years significant efforts among

many research groups have focused on enhancing laser-solid interactions, atomic

emission intensity, and detection limits with dual-pulse techniques [St-Onge 1998, Stratis

2000, Gautier 2005]. More specifically, Stratis [2000] used an orthogonal pre-ablation

spark to obtain 11 and 33-fold enhancement in spectral response of copper and lead,

respectively, in comparison to single laser LIBS. In their study, an orthogonal system was

used, in which a laser pulse was brought in parallel to the sample surface and focused a

few millimeters above it to form an air plasma or air spark. A few microseconds later a

second laser pulse, which was focused on the sample, traveled through the location of the

first plasma and ablated the sample material, forming the LIBS plasma from which

analyte emission was analyzed. Figure 1-3 illustrates the dual-pulse orthogonal









configuration along with another setup, which takes advantage of two in-line pulses.

Others to find similar enhancement results, some up to 40 times, include

SSecond Laser
Pulse Focused on
Second Laser Pulse the Sample
Focused on the
Sample Surface

First Laser-Induced
/ Plasma First Laser Pulse
First Laser Pulse Focused on the
Focused abow the Sample First & Second
Sample Surface Second Laser- Laser-Induced
Induced Plasma Plasma


Sample Sample

A B


Figure 1-3. Illustrations representing two setups used in many of the dual-pulse
studies; [A] An orthogonal configuration and [B] An in-line configuration.

Sattman [1995], Angel [2001], Coloa [2002], Scaffidi [2004], and Hohreiter [2005b].

Recent studies have examined the mechanisms for this dual-pulse enhancement, focusing

on the laser-solid coupling [Hohreiter 2005b, Linder 2005], as well as the hydrodynamics

of the overlapping laser pulses, shock waves, and subsequent density effects [Corsi

2004]. The exact mechanisms of signal enhancement with dual-pulse LIBS are complex

due to the combinations of laser ablation, analyte dissociation, and plasma excitation of

atomic species. Previous studies have shown that enhancements in atomic emission

intensity, peak-to-base, and signal-to-noise measurements with a dual-pulse system are

not simply the result of added energy from the pre-ablation laser-induced plasma. In fact,

dual-pulse signal enhancements are coupled to other physical phenomena such as

enhanced ablation, shock-induced plasma rarefaction (i.e., reduced density), or thermal

lensing from the pre-ablation plasma [Angel 2001, Coloa 2002, Hohreiter 2005b, Linder









2005]. While much has been learned from these studies of solids and liquids, the

extension of dual-pulse LIBS to gas-phase and aerosol systems brings along a new set of

dynamics which have not been explored prior to this study.

1.6 Objective

Laser induced breakdown spectroscopy is an up and coming quantitative aerosol

detection technique, but has problems competing with other analytical techniques,

especially when comparing detection limits. The objectives of this work are to present,

test, and compare possible solutions to increase analyte response in aerosol based LIBS,

consequently bettering limits of detection. A dual-pulse system and the use of a particle

focusing lens to introduce aerosols into the plasma were two such methods considered in

this work to enhance analyte response.

This study is the first to address a dual-pulse laser system for a gaseous and aerosol

application, noting that much research has been done on solid and liquid samples as

described above. Spectra and plasma transmission data from calcium aerosol samples and

air samples were collected and analyzed for both the single and dual laser cases.

Differences in measurements between the techniques and between the two different

samples were noticed, thereby helping to explain previously addressed plasma-particle

interactions.

The second part of the study focused on the idea that analyte response could be

enhanced when a particle lens was used to introduce aerosols directly into the laser

induced plasma. A particle lens aerodynamically focuses aerosol particles so that a tight

streamline of particles can be achieved, thus creating a more concentrated cross-section

of particles entering the plasma. Spectral response and hit rates of calcium aerosols were

compared when they were introduced through a particle lens versus a standard /4" thin






18


walled tubing. The following study will show that the LIBS technique can still be

improved from its current state.
















CHAPTER 2
EXPERIMENTAL METHODS

The current study was based on two experiments, LIBS improvements from a

dual-pulse system, and from introducing a particle focusing apparatus. As in all LIBS

configurations, the two experimental setups can be divided into (1) the plasma creation,

by means of focusing a laser source so that the threshold of the sample is reached, (2) the

collection of the spectral emission from the plasma with the use of a spectrometer and

ICCD, and (3) the analyte generation and introduction into the sample chamber.

2.1 LIBS Experimental Setup and Procedure

2.1.1 Plasma Creation Methods

High irradiance is needed to cause a breakdown in a gaseous material. This is done

by using a pulsed laser (10-20 ns duration and 50-300 mJ/pulse) and by focusing the

beam to reduce the cross sectional area. All the beam energy profiles followed a Gaussian

trend, peaking at the center of the beam and exponentially decreasing to its minimum at

the edges. Figure 2-1 shows a typical setup employed for plasma creation.




Aperture Plasma
Nd:YAG
laser
Condensing Lens
Gallilean Telescope

Figure 2-1. Optical setup for the delivery of the laser irradiance to the sample volume
resulting in the creation of a plasma.









Based on light focusing laws, the diffraction limit is inversely proportional to the incident

beam diameter, meaning that a tighter focus can be achieved when the beam diameter

approaching the condensing lens is large. For this reason, a Gallilean telescope is often

used to widen the laser beam prior to its final plasma creating focus. The Gallilean

telescope utilizes an expansion lens followed by a lens that collimates the expanding

beam. The aperture following the Gallilean telescope trims off the light on the edge of the

beam to allow for a more uniform beam profile before it reaches the focusing lens.

2.1.2 Emission Collection

Different configurations are used to direct the plasma emission onto the fiber optic

before it is processed by the spectrometer. In this study, a back collection method was

used. Due to variations in laser pulses and presence of aerosols, the location of the

plasma can fluctuate along the beam path ranging up to a millimeter or two. Rather than

collecting the light from the side, which would result in the fluctuating focused light

missing the entrance into the fiber optic, back collection was used so that the error from

the spatial variation in the plasma formation is minimized.

Once the plasma is created, equally intense light is emitted in every direction with a

portion of it being collimated by the same condensing lens used to focus the plasma

forming laser. The plasma light was then turned by a pierced mirror, allowing for the

laser beam to pass through and the emission to be reflected, and focused by a condensing

lens onto a fiber optic bundle leading into a spectrometer. The light was then dispersed by

the 0.275-m spectrometer (Acton SpectraPro-275) and finally recorded by a 256x1024

element intensified CCD array (Princeton Instruments model: 1024MLDS-E/1). The

spectrometer used a 2400 groove/mm grating, which provided an approximately 30-nm

wavelength range with 0.12 nm spectral resolution. The intensifier on a CCD served two












Pierced Mirror

Laser
Plasma < Spectro-
meter

Plasma Emitted Light ICCD




Fiber Optic


Figure 2-2. Optical setup for the collection of plasma emission from the sample
volume.

purposes, to allow for adjustment of signal gain and to allow for small time frames in

which light is being transmitted (gate width) to the CCD array. The latter is especially

important at small delays after the laser pulse, since the continuum light is so intense that

it would saturate the CCD camera if small gate widths were not applied. The laser Q-

switch is used to trigger a delay generator, which in turn triggers the ICCD, so that light

can be acquired at desirable delays after plasma creation.

2.1.3 Aerosol Generation

For all aerosol samples, a pneumatic-type medical nebulizer (Hudson model 1720)

was used to convert an aqueous solution into an aerosol spray. For all experiments, 5 1pm

of HEPA filtered air, regulated by a digital flow controller (Allicat Scientific), were used

to drive the nebulizer. The nebulizer sprayed the aerosol solution into a secondary flow of

air used to dry out the aerosol droplets and to transport them into the plasma sample

volume. Figure 2-3 shows a general schematic for the aerosol generation system. The co-

flow rates indicated in Figure 2-3 are those used in the dual-pulse study.










Plasma
n -. -1


Figure 2-3. General schematic of the aerosol generation system. The co-flow rates
were those used in the dual-pulse study [Hahn 2001].

Experiments were run to test the analyte concentration consistency in the nebulizer

spray. It was possible that what was exiting the nebulizer may not be at the same

concentration as the solution in the nebulizer due to the selective evaporation of the

matrix water. If the spray concentration was not the same as the liquid solution in the

nebulizer, then after elapsed running time, the concentration of analyte would be

enriched, growing linearly with elapsed time, as compared to the original solution. This

was tested by running the nebulizer for 10 and 20 minutes, both with a solution

containing copper analyte at the same initial concentration, and then comparing what was

left over inside the nebulizer to the original solution. The original solution, along with the











2500000


2000000


0 1500000
0


E 1000000


500000


0-
0 10 20
Nubulizer Run Time (min)

Figure 2-4. Chart displaying Cu analyte counts of remnant solutions from a nebulizer
as a function of nebulization time. Error bars represent the standard
deviation.

remnant solutions, were run through an ICP mass spectrometer, from which particle

counts of each sample were extracted. Figure 2-4 shows the average particle counts for

the original sample as well as the samples taken after 10 and 20 minutes. Each

measurement was done in triplicate, while each ICP-MS measurement was done in

duplicate.

Figure 2-4 shows no trend along with very little difference between all three

samples, indicating that the concentration in analyte was not changing and therefore the

concentration of the aerosol spray exiting the nebulizer was the same as the solution

inside the nebulizer over these time scales.










2.2 Dual-Pulse LIBS: Gas and Aerosol Study

2.2.1 Laser Configuration

Two Q-switched Nd:YAG lasers operating at their fundamental wavelength of

1064 nm and at a repetition rate of 5 Hz were used in all measurements for this

experiment. A schematic of the optical configuration is shown in Figure 2-5. The first



Plasma Volume Pierced Mirror



Laser 2
Nd:YAG
Power Meter
(Transmission
Measurments) Laser I

Fiber
Optic



Spectro- CCD
meter



Figure 2-5. The experimental apparatus for a single and dual-pulse LIBS
configuration.

laser (Continuum Precision II 8000), which will be referred to as Laser 1, used a laser

beam energy of 100 mJ/pulse, and was focused to the center of the sample chamber (a

six-way vacuum cross) using a 100-mm focal length UV grade plano-convex lens. The

pulse-to-pulse stability was directly measured (Ophir Nova II), which yielded a relative

standard deviation (RSD) of 0.42%, with a maximum pulse deviation of 1.4%. The pulse

energy was sufficient to create a laser-induced plasma with Laser 1 operating alone in

either a purely gas-phase or aerosol laden sample stream. The second laser (Big Sky CFR

400), which will be referred to as Laser 2, operated at a laser pulse energy of 290









mJ/pulse, and was focused with a 75-mm focal length UV grade plano-convex lens, also

creating a laser-induced breakdown in the center of the six-way cross. The pulse-to-pulse

stability was (RSD) of 0.45%, with a maximum pulse deviation of 1.5%.

The two laser beams were carefully aligned such that the two plasmas were formed

at the identical spatial location (see below for alignment details). No quantitative plasma

volume measurements were recorded. However, the Laser 2 plasma was previously

characterized in detail; hence a characteristic plasma diameter is reported as 1.5 mm

[Carranza and Hahn 2002b]. Furthermore, the Laser 2 plasma was noticeably larger than

the Laser 1 plasma, presumably due to the greater laser pulse energy.

As seen in Figure 2-6, the flashlamp sync of Laser 1 was used to trigger a digital

delay generator, which was then subsequently used to trigger the flashlamp and Q-switch

of Laser 2. Adjustment of the delay generator allowed for the two laser pulses to fire

simultaneously, or at variable pulse-to-pulse delays ranging up to 1 ms, the largest

investigated. For all experiments, the delay was adjusted such that Laser 1 was fired first,

followed by Laser 2 at the specified delay. The relative flashlamp to Q-switch timing was

maintained constant for both lasers for all experiments, thereby ensuring constant laser

pulse energy and laser beam characteristics. It is noted that a delay time of zero

corresponds to both Laser 1 and Laser 2 pulses firing simultaneously. A fast response

(200-ps rise time) detector and digital oscilloscope (2.5 Gsample/s) were used to

continuously measure and monitor the temporal delay between the two laser pulses for all

experiments. As implemented, the jitter between Laser 1 and Laser 2 was typically less

than 5 ns.










Laser 1
Laser 1 pulse
Flashlamp 277.75 [s Fixed To Delay
Sync Generator 1

77.75 [s + ,
Delay variable delay To Delay
Generator 1 Generator 2



Delay To Laser 2
Generator 2A I Flashlamp


Delay 200 ps Fixed To Laser 2
Delay
Generator 2B Q-Switch

Laser 2
pulse
Figure 2-6. Schematic representing the trigger setup to achieve laser pulses fired
simultaneously in time or at a variable delay. Delay generator 1 set to
77.75 |ts would result in the lasers firing simultaneously. Adjusting the
variable delay of generator 1 would result in delaying laser 2 pulse.

2.2.2 Experimental Analyte 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 [Hahn and Lunden 2000, Hahn 2001]. As seen in Figure 2-3, the purely

gaseous sample stream consisted of 40 1pm of purified air, which 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 (Matheson model: 8270).

For the aerosol measurements, two types of calcium-containing particles were used,

with the purpose of studying any potential size effects. The majority of aerosol

measurements were made by nebulizing a solution of 50 [g Ca/ml at a rate of about 0.1

ml/min. The nebulizer output was introduced in a gaseous co-flow stream of 40 1pm of









purified air. The calcium solutions were prepared by diluting ICP-grade analytical

standards of 10,000 |tg Ca/ml (SPEX CertiPrep). Accounting for the additional 5 1pm

used to drive the nebulizer, this configuration resulted in a calcium-rich aerosol flow of

approximately 100 |tg Ca/m3 through the LIBS sample chamber. Based on previous TEM

measurements using the current configuration [Hahn 2001], 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. The corresponding

particle number density is on the order of 105 cm-3 in the LIBS sample chamber, which

yields an average number of analyte particles per plasma volume on the order of 100.

Overall, the system provides a highly dispersed, submicron-sized calcium-rich aerosol

stream for LIBS analysis. This analyte source will be referred to as the fine calcium

aerosol experiments.

In addition, some experiments were performed by nebulizing a suspension of

nominally 2-[tm sized ( 0.7 [tm) borosilicate glass microspheres (Duke Scientific,

#9002) in ultrapurified water. Based on previous analysis, the calcium concentration

within the glass microspheres was determined to be about 2% (by mass), which yields a

strong calcium atomic emission signal [Hohreiter 2006]. The particle concentration in

suspension was adjusted so that the resulting borosilicate particle number density

produced in the LIBS sample chamber was on the order of 102 cm-3. These experimental

conditions will be referred to as the borosilicate microsphere experiments.

2.2.3 Spectral Measurements

Both laser-induced plasmas were first visually aligned to the same spot, noting that

the Laser 2 plasma emission was previously aligned to the fiber optic such that both









atomic and continuum plasma emission were optimized. Once the Laser 1 plasma was

visually centered on the Laser 2 plasma, the final alignment of Laser 1 was performed to

maximize plasma emission (atomic and continuum emission) coupling to the fiber optic,

thereby ensuring alignment of both laser-induced plasmas to the same spatial location.

For all single-pulse and dual-pulse experiments, the external Q-switch sync from

Laser 2 was used to trigger the ICCD controller. Hence for a given set of experiments, the

ICCD was fixed relative to the temporal position of Laser 2. An additional delay was then

introduced between Laser 2 and the ICCD detector gate, which allowed for optimization

of the specific analyte atomic emission signals. For the purely gas-phase experiments

(i.e., nitrogen and oxygen atomic emission analysis), an ICCD gate delay and gate width

of 5 |ts were used, therefore spectral integration was initiated 5 |ts following Laser 2. For

the aerosol experiments (i.e., calcium atomic emission analysis), both the detector delay

and the gate width were increased to 30 [ts.

Spectral data was acquired using an ensemble average of 1000 laser shots. The

process was repeated a total of three times for each different analyte, and for each

adjusted dual-pulse delay. In addition to dual-pulse measurements, single-pulse data were

recorded for both Laser 1 alone and Laser 2 alone for all different analytes. Finally,

single-particle analysis was used for the 2-[tm sized borosilicate glass particles, as

reported in previous studies [Hahn and Lunden 2000, Hohreiter 2006]. These single-shot

experiments were performed using 500 shot sequences, repeated three times for each

experimental condition.









2.2.4 Transmission Measurements

For the transmission experiments, laser pulse energy measurements were made for

Laser 2 at a spatial location directly in front of the sample chamber (i.e., incident energy)

and at a location directly exiting the sample chamber (i.e., transmitted energy). This was

achieved by placing the laser energy meter (Ophir Nova II) in front of the six-way cross

to measure the incident beam energy, and immediately after the chamber to measure the

transmitted laser beam energy, with the latter configuration shown in Figure 2-5. The

average transmitted energy (i.e., Laser 2 transmission) was then calculated from the

direct ratio of these two measurements.

The transmission measurements were recorded for both single-pulse (Laser 2 only)

and dual-pulse (Laser 2 following Laser 1) configurations, for both the gaseous and

aerosol systems, including a full range of dual-pulse, laser-laser delay times. An

ensemble-average of 500 laser shots was recorded for both the incident and transmitted

pulse energies, which were repeated a minimum of three times each. In addition to the

mean energy values, full statistical parameters were recorded, which included the

minimum, maximum, and standard deviations of the shot-to-shot pulse energies.

2.3 Particle Lens Study

2.3.1 Laser Configuration

An Nd:YAG laser (Continuum Surelite II) with a repetition rate of 5 Hz was run

through two wavelength doubling crystals yielding a 355 nm plasma creating laser

source. As seen in Figure 2-7, the laser source was turned by a series of 355 nm dichroic

mirrors with anti-reflection coatings at 532 and 1064 nm to eliminate any excess

fundamental 1064 nm and green 532 nm light. The beam energy measured after the 355

nm mirrors just before reaching the condensing lens was about 60 mJ/pulse. The laser









beam passed through a pierced mirror and was finally focused by a 100 mm focal length

UV grade plano-convex lens to create the plasma.

Fiber Optic


Aerosol and Air
355 ARMixture

mirrors

Nd:YAGr / Particle Lens

laser



Pierced
Mirror
Vacuum Pumped Out


Figure 2-7. The optical setup (top view) for the particle lens experiments. Like the
dual-pulse setup, the plasma emission was back collected. The aerosols
were drawn through the particle lens into the plasma by a vacuum pump
situated on the opposite side of the 5-way cross.

2.4.2 Analyte Generation

For all measurements, the aerosols were created by nebulizing a solution of

nominally 2-[tm sized ( 0.7 [tm) borosilicate glass microspheres (Duke Scientific,

#9002) in ultrapurified water, the same used for the borosilicate experiments in the dual-

pulse study. Accounting for 5 1pm used to drive the nebulizer and the 5 1pm of co-flow

used to dry the nebulized droplets, a flow was created yielding approximately 30 tg

Ca/m3. A portion of the nebulized flow, about 1 lpm (based on the optimal conditions

determined by the manufacturer, Aerodyne Research, Inc.) was pulled through the

particle lens and the plasma chamber by a vacuum pump. The flow rate through the

vacuum pump was regulated by a rotameter-type flowmeter (Gilmont Instruments'). To










determine the particle lens' effectiveness, a standard thin walled steel tube with an inner

diameter of about 0.18" was also used in all experiments to transport the aerosols to the

plasma.

The particle lens was developed by Aerodyne Research, Inc. and consisted of a

series of concentric tubes that stepwise narrowed the flow down to a cross section with a

diameter of 1.65 mm before releasing the streamlined particle flow out of a 0.75 mm

orifice port. Figure 2-8 shows a general schematic of the particle lens.

Annular co-flow
used to isolate
particle stream

Aerosol i Plas ma


Particle lens outlet was visually aligns

0.75 mm
diameter


Figure 2-8. Simple schematic of the particle lens. Aerosols flow through a series of
radial decreasing concentric tubes before exiting through an outlet causing
a streamline exit into the laser induced plasma

The particle lens outlet was visually aligned vertically and horizontally to the laser

induced plasma and was situated about 2 mm away. At first attempts, it was determined

that the particles were filling and re-circulating throughout the entire sample chamber

(the 5 way cross), therefore rendering the particle lens useless, since there was no

preferential increase in particle concentration at the exit of the particle lens. To remedy

the recirculation problem, a secondary sheath flow was introduced annularly around the

particle lens. This was accomplished by placing a 1/2" thin walled steel tube around the

particle lens and the 14"tube, represented by the outer lines in Figure 2-8. Air was flown









through the 1/2" tube at a flow rate of 9 1pm and was shown to be effective in separating

the particles exiting the lens with those re-circulating in the chamber.

2.4.3 Spectral Measurements

Similar to the other study, the resulting emission was back collected, turned by the

pierced mirror and focused by a 100 mm focal length lens onto a fiber optic bundle. The

fiber optics fed into the same spectrometer and ICCD that was used in the spectral

collection for the previous study. The ICCD controller was triggered by the Nd:YAG

laser's external Q-switch, initiated 20 |ts after the laser pulse with a 30 |ts gate width.

Analyte response was compared using spectral data acquired through an ensemble

average of 1000 laser shots for each type of aerosol injection tube (the particle lens and

the 1/4" thin walled tube). A conditional analysis approach, mentioned in Chapter 1, was

also used to gather hit rates and to acquire individual spectra that could be used to further

investigate any advantages in using a particle lens.

2.4.4 Particle Counting Measurements

To help quantify the difference in hit rates and analyte response between the two

injection methods, the transmission of particles through both tubes was calculated. This

was done by flowing the analyte through the /4" tube and the particle lens both at 1 1pm.

A particle counter was placed at the end of the 5-way sample chamber, which drew out

0.28 1pm from the exiting 10.28 1pm flow (combination of the sheath flow, 9.28 1pm, and

the analyte co-flow mixture, 1 1pm). The particle counter would calculate the number of

particles between a certain size range. Since the experimental analyte nominal diameter

was 2.2 [am, all counts of particles above 2 [am were considered. The particle counter

used was a Lasair model 1001 (Particle Measuring Systems Inc.) that used light scattering

to determine particle number and size.
















CHAPTER 3
RESULTS AND DISCUSSION

3.1 Dual-Pulse LIBS Study

This study focused on the effects of an orthogonal dual-pulse laser configuration on

the atomic emission response for both purely gaseous and calcium-based aerosol samples.

The purpose was to determine if a dual-pulse configuration, which has been shown to

increase analyte response in solids and liquids, could improve a gas and aerosol analyte

response over that of a typical single-pulse LIBS system. Transmission measurements

were made to help physically explain the differences observed in analyte response

between the two configurations and between the gas and aerosol analyte.

3.1.1 Transmission Experiments

Figure 3-1 presents the transmission of Laser 2 as a function of delay time between

Laser 1 and Laser 2 for both the pure air stream and the fine calcium aerosol experiments.

As a reference, the dashed line in Figure 3-1 represents the average transmission of

43.8% for the single-pulse configuration (i.e., Laser 2 only), which is the average of the

purified air (44.5% with 2.7% RSD) and the fine calcium particle (43.1% with 0.8%

RSD) transmission values. The overall behavior of the transmission data as observed in

Figure 3-1 is rather complex, and is considered for discussion purposes in terms of four

distinct temporal regions. The average RSD values for the dual-pulse transmission

measurements were 4.4% for the pure air stream and 3.3% for the fine calcium particles,

as averaged over all temporal delays. The corresponding error bars where comparable to










the symbol size for most values, hence error bars were omitted in Figure 3-1 to avoid

clutter, given the partial overlap of many symbols.


A Pure Air
v Calcium Aerosol
















V








I I, I i I I ,


AAA


v V

V
A




A





V





, 1, ,,,,, I ,,,I ,, ,,I ,,


0 20 40 60 80 100 103 104 105 106
Delay Time (ns)

Figure 3-1. Laser 2 transmission as a function of dual-pulse laser delay times for the
pure air and fine calcium aerosol sample streams. The horizontal line
represents the transmission of Laser 2 alone (i.e., single pulse LIBS). Note
the plot is linear for delay times less than 100 ns.

The first temporal region to be considered will correspond to dual-pulse delays

less than 100 ns. During this temporal range, there was little transmission of Laser 2

through the plasma formed by Laser 1, as explained below. The minimum recorded

transmission was 3.3% (1.3% RSD) and 3.1% (8.2% RSD), which both occurred at a

delay of 25 ns, for the gas-phase and fine calcium aerosol phase, respectively. At these









delays, the plasma formed by Laser 1 is essentially opaque to the incident Laser 2 pulse,

which follows a previously reported trend for the temporal characteristics of laser-

induced plasmas [Hohreiter 2004b]. During this temporal region, the high plasma

temperatures and free electron densities (-1018 cm-3) of the first plasma result in an

optically dense plasma for the Laser 2 incident radiation. The resulting highly energetic

plasma state is essentially independent of the presence of the calcium aerosol, therefore

no difference is observed between the two experimental conditions. This result is

consistent with previous measurements, in which identical plasma temperatures and free

electron densities were recorded for gas-phase and particle-seeded flows under conditions

similar to the present study [Hohreiter 2005a], and no effect of aerosol presence on the

temporal location of plasma initiation was recorded [Hohreiter 2004a]. Finally, it is noted

that qualitative observations of the plasma revealed no differences between the pure air

and the aerosol seeded conditions, with regard to plasma size and spatial stability.

The second temporal region of interest in Figure 3-1 is considered to range from a

delay of 100 ns to about 1 hts. During this region, the highly absorptive nature of the first

plasma decreases as the free electron density and temperature rapidly decrease. Once

again, this is consistent with previous measurements, where an essentially identical laser-

induced plasma was found to be nearly transparent to a low-energy probe beam by about

500 ns following plasma initiation [Hohreiter et al 2004b]. Overall, as the delay between

Laser 1 and Laser 2 nears about 1 hts, the effect of the Laser 1 plasma is not very

significant regarding the coupling of Laser 2 energy into the existing Laser 1 plasma, as

the overall transmission of Laser 2 is near its single-pulse value. In other words, with the

dual-pulse configuration, Laser 2 initiates a breakdown process comparable to what









would occur in the presence of Laser 2 alone, resulting in a similar coupling of the

incident energy into the resulting plasma.

The third temporal region corresponds to delays ranging from 1 |ts to about 100

hts, and is defined by a relatively high transmission of the Laser 2 pulse. Specifically, the

transmission of Laser 2 has a maximum value of 90.0% (0.82% RSD) and 79.5% (1.1%

RSD) for the pure air and fine calcium aerosol systems, respectively, over this temporal

region. Unlike in the earlier temporal regions, an effect of the fine calcium aerosol is now

observed, as the transmission is slightly reduced in the presence of aerosol in comparison

to the pure gaseous system. This reduction is indicative of some laser-particle interactions

during the breakdown event, as discussed in more detail below.

Finally, the fourth temporal region corresponds to laser-laser delays from 100 |ts

up to 1 ms. At the latter value, the transmission of the Laser 2 beam is observed to

approach the average single-pulse (i.e., Laser 2 only) value. Clearly at such large delays,

the Laser 1 plasma has sufficiently decayed in both temperature and free electron density

such that its effects on the subsequent laser pulse (i.e., Laser 2) are negligible.

The transmission behavior of the first two regions is not unexpected, given the

earlier studies on temporal plasma characteristics; hence a strong plasma-laser interaction

during this period drives the dual-laser coupling and is virtually independent of the

presence of particulates. Furthermore, the behavior of the last region is expected as well,

given that at significantly long laser-laser delays, the two laser pulses must approach

independence with regard to interaction. Clearly -1 ms is nearing this asymptotic limit

for pulse-to-pulse independence. Given these comments, a key region of interest with

regard to the physics of dual-pulse LIBS for gaseous and aerosol systems is the third









region, namely between about 1 to 100 |ts of laser-laser delay. It is well known that the

laser-induced plasma is characterized by a rapid plasma expansion and a concomitant

shock wave. Not withstanding the increased pressure behind the shock front, the

significant plasma temperature results in a reduced density (i.e., mass/volume) within the

resulting plasma. This rarefaction is important with regard to the coupling of Laser 2 into

the existing Laser 1 plasma, as related to the breakdown threshold.

At ambient temperature, the laser-induced breakdown threshold is known to vary

inversely as pressure, ideally as p-12, noting that both multiphoton and cascade ionization

processes are important in plasma formation and growth. Therefore, the ionized plasma

from Laser 1, albeit at reduced density, presents a more complex problem for treatment of

the Laser 2 interaction than would be predicted from treatment of pressure/density effects

alone. However, the significant decrease in Laser 2 energy coupling into the plasma (i.e.,

increased transmission) is interpreted in terms of an increased breakdown threshold. This

increased threshold effectively delays the temporal "breakdown" point of the Laser 2

pulse to later in the pulse waveform, thereby resulting in less coupling of Laser 2 pulse

energy. It is well known that the presence of aerosol particles can considerably lower the

breakdown threshold [Lencioni 1972 and Smith 1997]. Therefore, the slight reduction in

transmission between 5 and 20 |ts delay from the pure air stream (c = 89.7% with 0.7%

RSD) as compared to the fine aerosol stream (c = 75.0% with 1.5% RSD) is consistent

with the addition of the calcium-rich particulates and a concomitant reduction in

breakdown threshold.









3.1.2 Spectral Analysis of Gaseous Analyte

Neutral atomic nitrogen (N I) lines at 491.4 nm and 493.5 nm (86,137 106,178

cm-1 and 86,221 106,478 cm-1, respectively), and the oxygen triplet centered at 394.7

nm (73,768 99,094 cm-) were used for the spectral analysis of the purified air sample

stream. As an example, nitrogen spectra for the single-pulse (Laser 2 only) and the dual-

pulse (500 ns delay) configurations are presented in Figure 3-2.


12000



10000



8000


6000


4000 V -r
Dual Pulse v"^

2000




480 485 490 495 500
Wavelength (nm)

Figure 3-2. Spectra showing the two nitrogen atomic emission lines at 491.4 nm and
493.5 nm. The spectra correspond to dual-pulse LIBS with 500 ns delay
(lower spectrum) and single-pulse (Laser 2 only) LIBS (upper spectrum).
Both spectra have the same scale.

Recalling that the ICCD is also synchronized to Laser 2, it is observed that the

overall signal intensity is somewhat reduced by the dual-pulse scheme for this temporal









regime. However, at significantly short laser-laser delay times (~0 to 100 ns), the dual-

pulse configuration produced a slightly greater emission intensity than Laser 2 alone,

which is consistent with the enhanced coupling efficiency of Laser 2 as observed in the

transmission experiments.

To quantify the emission signals, both the peak-to-base and signal-to-noise ratios

were calculated using the 493.5-nm N I spectral line for both the dual-pulse and the

single-pulse configurations. The peak-to-base is perhaps the most widely used LIBS

signal metric, as it provides a normalization of the atomic emission line with the plasma

continuum emission, allowing for a more precise emission signal as noted previously

[Coloa 2002 and Carranza and Hahn 2002a]. As an analytical figure of merit, the signal-

to-noise ratio is the more relevant metric. The signal-to-noise ratios (SNR) were

calculated from the integrated peak intensities and the calculated root-mean-square noise

from the adjacent continuum region. Figures 3-3 and 3-4 present the peak-to-base ratio

(P/B) and signal-to-noise ratio as a function of dual-pulse delay for the 493.5-nm nitrogen

line. Before discussing the figures, it is important to note that Laser 1 alone produces a

significant laser-induced plasma; hence analyte emission is observed over a range of

detector gate delays stemming from this emission source only. Therefore, as the laser-

laser delay is being adjusted for dual-pulse experiments, the effective detector gate delay

with respect to Laser 1 is also being varied, noting that the detector gate is fixed with

respect to Laser 2. It is therefore important to consider the analyte signals stemming from

Laser 1 alone (i.e., Laser 2 blocked), which are therefore included in Figures 3-3 and 3-4.







40


28
A Laser 1 Only
V Dual Pulse


26
Z
E
IF
CO
I 24 A









20
M -- -- -- -- -- -- -- --X- -- -
18





0 200 400 600 800 1000
20




18 ---------
0 200 400 600 800 1000
Delay Time (ns)

Figure 3-3. The 493.5-nm nitrogen emission line peak-to-base ratio measurements for
the pure air sample as a function of dual-pulse laser delay times. The P/B
ratios are also shown corresponding to the Laser 1 plasma only (i.e., Laser
2 beam blocked). The dashed horizontal line represents the average P/B
ratio for Laser 2 only. A representative error bar is included on the 100 ns
Laser 1 only mark.

Figure 3-3 shows the dual-pulse results as compared to the single-pulse LIBS

configuration (i.e., Laser 2 only), which is indicated with the dashed horizontal line. For

the range of delays from 0 to 1000 ns, the P/B ratios are observed to depart little from the

single-pulse average of 22.8 (0.21% RSD). The maximum dual-pulse value occurs at 250

ns delay, and corresponds to a P/B ratio of 23.8 (0.41% RSD), or a 4% increase with the

dual-pulse configuration. While statistically significant, a 4% signal enhancement would

not justify the increased experimental complexity and expense of the dual-pulse scheme.

By 1 |ts delay time, the dual-pulse P/B ratio has decreased to 21.3 (0.47% RSD), or a







41


reduction of 6.8% with regard to single-pulse LIBS. Data were collected for larger laser-

laser delay times, but this trend of decreasing response continued, with a dual-pulse P/B

ratio of only 9.3 (0.50% RSD) recorded at 5 |ts delay. This corresponds to a 60%

decrease in analyte signal with the dual-pulse scheme at this larger delay, as a result of

the diminished Laser 2 coupling consistent with Figure 3-1.



70
A Laser 1 Only
V Dual Pulse
60

Z
c 50
L?





0


S20 -
30


0





0 200 400 600 800 1000
Delay Time (ns)

Figure 3-4. The 493.5-nm nitrogen emission line signal-to-noise ratio measurements
for the pure air sample as a function of dual-pulse laser delay times. The
SNR are also shown corresponding to the Laser 1 plasma only (i.e., Laser
2 beam blocked). The dashed horizontal line represents the average SNR
ratio for Laser 2 only.

The SNR values, as shown in Figure 3-4, display a similar trend to the P/B ratios,

although the maximum enhancement for the dual-pulse configuration is shifted from 250

ns delay to a delay of 500 ns. Specifically, the SNR is increased from a single-pulse value










of 33.0 (0.47% RSD) to a dual-pulse value of 47.6 (2.4% RSD) at 500 ns, which

corresponds to a 44% improvement. Once again, by a 5 |ts laser-laser delay, the dual-

pulse SNR was decreased by 86%, to a value of only 4.6 (0.10% RSD).

The oxygen atomic emission lines revealed similar, although not as positive,

trends as the nitrogen data. Figure 3-5 displays the 394.7 nm oxygen spectral emission

line for the 500 ns laser-laser delay dual-pulse case as well as the spectra for the single

laser 2 only case. The P/B ratios were calculated using the 394.7-nm O I triplet and can


10000




8000


6000


c

_ 4000




2000




0
376


380 384 388 392
Wavelength (nm)


396 400 404


Spectra showing the oxygen (I) triplet at 394.7 nm. The spectra
corresponds to dual-pulse LIBS with 500 ns delay and single-pulse (Laser
2 only) LIBS. Both spectra have the same scale.


Figure 3-5.







43


be seen in Figure 3-6. For a laser-laser delay of 250 ns, the P/B ratio was decreased from

a single-pulse value of 18.1 (0.79% RSD) to a dual-pulse value of 17.0 (0.40% RSD), or

a decrease of 6.5%. By 5 hts, the dual-pulse P/B ratio was reduced to 4.2 (8.1% RSD),

corresponding to a reduction of 76.9%. A signal-to-noise plot was also constructed for

the 394.7 nm oxygen spectral emission line, which can be seen in Figure 3-7. The trend

of the S/N plot followed similarly to the oxygen P/B plot. Like the P/B plot, the signal-to-

noise ratios indicate a decrease in analyte response for the dual-pulse configuration when

compared to the single Laser 2 only (represented by the dashed line) case.


400 600
Delay Time (ns)


Figure 3-6.


1000


Oxygen peak-to-base measurements for the filtered air sample as a
function of dual-pulse laser delay times. The average peak-to-base ratio
for Laser 2 only is represented by the dashed horizontal line.


A Laser 1 Only
V Dual Pulse




A
A



A


SA-







44


45



40






S30


25
CO
n-~











15 -
3A
S20 -





V Dual-Pulse S/N
A Laser 1 Only S/N
10
1 0 I I... I I I I I I I I I I I I

0 200 400 600 800 1000
Delay Time (ns)


Figure 3-7. The 397.4-nm oxygen emission line signal-to-noise ratio measurements
for the pure air sample as a function of dual-pulse laser delay times. The
S/N are also shown corresponding to the Laser 1 plasma only (i.e., Laser 2
beam blocked). The dashed horizontal line represents the average S/N
ratio for Laser 2 only.

The above results appear at first glance to contrast with much of the current

literature on dual-pulse LIBS, which tends to show significant signal enhancements with

dual-pulse schemes. However, the mechanisms and physics of purely gas-phase LIBS

analysis are fundamentally different from the analysis of bulk solids or liquids. With the

latter, the reduced density behind the shock wave of the first plasma enhances the second

laser's interaction with the solid sample, while simultaneously reducing the plasma

coupling with the gaseous matrix; thereby increasing the analyte response of the solid. In

contrast, the present results must be explained in the context of the transmission









experiments, as well as in terms of recent experiments comparing the response of gas-

phase and particulate phase analytes [Hohreiter 2005a].

For the relatively short laser-laser delay times (Z 1 Its), the Laser 2 pulse is

efficiently coupled to the Laser 1 plasma, leading to a greater quantity of energy coupled

into the resulting dual-pulse plasma. Increased total laser pulse energy into a laser-

induced plasma event may or may not improve the analyte signal response, as explored in

detail in an earlier study [Carranza and Hahn 2002a], depending on the overall plasma

regime. In the current study, the additional energy with the dual-pulse configuration

produced a marginal increase in nitrogen emission, and a marginal decrease in oxygen

emission, during this delay period. This trend most likely reflects the slightly different

upper energy states of the nitrogen and oxygen transitions, which are expected to

correspond to slightly different optimal temporal windows, as based on plasma

temperature [Fisher 2001]. By coupling more energy into the resulting dual-pulse plasma,

the temporal temperature profile of the plasma is altered, thereby altering the peak-to-

continuum emission characteristics at the fixed detector gate. This behavior is further

reflected by comparing the Laser 1 only data to the Laser 2 only data at zero delay time,

which has the effect of changing the laser pulse energy from 100 mJ to 290 mJ with a

fixed detector gate. The nitrogen P/B ratio is observed to increase by 3% with increased

pulse energy, while the oxygen P/B ratio is observed to decrease by 5%. Clearly the

temporal location of optimal P/B response is different for the selected nitrogen and

oxygen emission lines.

Notwithstanding the above comments, for the longer laser-laser delay times

(beyond 1 [ts), a very different dynamic is observed for both gaseous analyte species.









Specifically, for this increased delay regime, the dual-pulse configuration yielded a

significant decrease in analyte response, which again is explained in the context of Figure

3-1, and the generally lessened LIBS analyte response with gas-phase species as recently

reported [Hohreiter and Hahn 2005a]. In their study, Hohreiter and Hahn proposed that

the expanding shockwave of the laser-induced plasma would preferentially expel (i.e.,

push) molecules from within the plasma core, producing a decreased concentration of

analyte for gas-phase species, represented in Figure 3-8.




*^ *q *

W. elk 6 l e:










their concentration in the plasma core [Hohreiter and Hahn 2005a].












emission data. Stated another way, the dual-pulse advantage of de-coupling the gas-phase

matrix from the bulk analyte phase, as realized with solid and liquid analysis, is not
a.. *, -.










possible with a dual-pulse analysis of a purely gas-phase system, because the gas-phase

matrix itself is the actual analyte of interest.

3.1.3 Spectral Analysis of Fine Calcium-Based Aerosol Analyte

In contrast to the above gas-phase analyte results, the analysis of the aerosol sample

streams affords an opportunity to realize the benefits of dual-pulse LIBS by attempting to


2500




2000




S1500
C


S1000




500


0 I
385


390 395


Wavelength (nm)


Figure 3-9. Spectra showing the Ca II atomic emission lines at 393.4 and 396.9 nm for
both the dual-pulse configuration with a 250-ns delay, and for Laser 2
only. Both spectra have the same intensity scale, and the dual-pulse
spectrum has been shifted upward by 400 counts for clarity.

decouple the particulate-phase derived analyte from the gas-phase species. With the

addition of aerosol particles into the sample stream, distinctly different results as










compared to the gas-phase experiments were observed. For these experiments, the first

ionized calcium (Ca II) atomic emission lines at 393.4 and 396.9 nm (0-25,414 cm-1 and

0-25,192 cm-1, respectively) were used for all spectral measurements due to their strong

intensity. As an example, recorded spectra are presented in Figures 3-9, 3-10, and 3-11,

2500




2000




S1500


U6

C 1000


Dual Pulse
500

Laser 2 Only



385 390 395 400
Wavelength (nm)

Figure 3-10. Spectra showing the Ca II atomic emission lines at 393.4 and 396.9 nm for
both the dual-pulse configuration with a 750-ns delay, and for Laser 2
only. Both spectra have the same intensity scale, and the dual-pulse
spectrum has been shifted upward by 400 counts for clarity.

as acquired using the dual-pulse configuration with a laser-laser delay of 250 ns, 750 ns,

and 50 [is, respectively. Along with the dual-pulse spectra, for comparison, spectra from

the single-pulse (Laser 2 only) configuration is also presented in each of the mentioned

figures. Several features are noted in Figure 3-10, which represents the 750 ns laser-laser

delay. The relative intensity of the Ca II atomic emission peaks is significantly greater for








the dual-pulse method as compared to the single-pulse method for, namely 1980 counts

vs. 1300 counts for the 393.4 nm peak, respectively. In addition, the region of the spectra

to the left of 393.4-nm Ca II line corresponds to molecular emission from the N first

negative system, including the lines at 391.4 and 388.4 nm. In contrast to the particle-

1500


1000


500






0


Lase 2 Only

' Dual-Pulse


Wavelength (nm)

Figure 3-11. Spectra showing the Ca II atomic emission lines at 393.4 and 396.9 nm for
both the dual-pulse configuration with a 50-ps delay, and for Laser 2 only.
Both spectra have the same intensity scale.
derived calcium emission lines, these gas-phase derived molecular lines display an

opposite trend, whereby the emission intensities are significantly reduced with the dual-

pulse configuration. This latter trend is perfectly consistent with the above analysis of the

gas-phase atomic nitrogen and oxygen emission data.


I_;;~;;










In a similar manner as for the gas-phase experiments (noting the longer detector

gate and width of 30 [ts), comparisons of Ca II atomic emission peak-to-base and signal-

to-noise ratios were made for the single-pulse and dual-pulse configurations, with the

results presented in Figures 3-12 and 3-13, respectively.

600
A Laser 1 Only
v Dual Pulse

500
o



CO


io 300 T
Co




100


0 0
(r








0 200 400 600 800 1000 104 105
Delay Time (ns)
Figure 3-12. The 393.4 -nm calcium II emission line peak-to-base ratio measurements
for the fine calcium aerosol sample as a function of dual-pulse laser delay
times. The P/B are also shown corresponding to the Laser 1 plasma only
(i.e., Laser 2 beam blocked). The dashed horizontal line represents the
average P/B ratio for Laser 2 only.

The dual-pulse measurements for both the peak-to-base and signal-to-noise are

significantly greater than those realized with the single-pulse scheme over a range of

laser-laser delay times from about 100 ns to nearly 100 Its. Specifically, the P/B ratio for

the single-pulse experiments (Laser 2 only) was an average of 220 (7.0% RSD). With the






51


dual-pulse configuration, the P/B was observed to increase to values of 460 (6.7% RSD)

at 750 ns delay, and 463 (4.6% RSD) at a delay of 5 hps. Nearly identical trends were

observed with the SNR data, as seen in Figure 3-13. The average single-pulse (Laser 2

only) SNR value was 43.4 (7.1% RSD), which was increased to a maximum value of


250
A Laser 1 Only
V Dual Pulse


o 200 T
E


C)
0
co 150


r
(C)
o 100
Z


c)
0 50


( 50 i I _ J1_I__


0 200 400 600 800 1000 104 105
Delay Time (ns)

Figure 3-13. The 393.4 -nm calcium II emission line signal-to-noise ratio
measurements for the fine calcium aerosol sample as a function of dual-
pulse laser delay times. The SNR are also shown corresponding to the
Laser 1 plasma only (i.e., Laser 2 beam blocked). The dashed horizontal
line represents the average SNR for Laser 2 only.

202 (10.5% RSD) for a dual-pulse configuration with a laser-laser delay of 5 hps. This

corresponds to a greater than 4-fold increase in analyte sensitivity with the optimal dual-









pulse scheme, which corresponds to a laser-laser delay in the range of about 0.8 to 5 [ts.

It should also be noted that the Laser 1 (single laser only) P/B and SNR data are all

markedly less than the dual-pulse results. The Laser 1 response corroborates the dual-

pulse enhancement, and results from a steady decrease in calcium emission response with

increasing laser-to-detector delay time, as effectively realized with the Laser 1 only

experiments. Recall that with Laser 1 only, the detector gate is still being temporally

delayed even though the Laser 2 beam is blocked.

As discussed above with the purely gas-phase data, the aerosol-phase experiments

should also be explained in the context of the transmission experiments and the recent

gas-phase vs. particulate-phase analyte response study [Hohreiter and Hahn 2005a].

Referring once again to this earlier study, the solid particulates (i.e., aerosol phase

analyte) were hypothesized to resist being driven from the plasma center along with the

gaseous molecules by the expanding shock wave, as seen in Figure 3-14. The resulting




0 *




*
r* 0
*a *


Figure 3-14. This image depicts the basis for the increase analyte response in a dual-
pulse configuration for an aerosol sample. The larger aerosol particles
resist the radial force exerted on them by the expanding plasma volume
created by laser 1 and remain inside the core while the smaller gas phase
particle are expelled, thereby creating a more aerosol concentrated volume
that is awaiting the plasma from laser 2 [Hohreiter and Hahn 2005a].









"slip factor" has the effect of preferentially enhancing the particulate-phase/gas-phase

analyte ratio, thereby affording an opportunity to enhance the analyte response of the

solid phase with an optimally-timed second laser pulse. In other words, the gas-phase

species within the plasma effectively contribute to the plasma continuum emission, hence

to spectral noise with regard to the targeted particulate-derived analyte emission. In

keeping with the Figure 3-1 transmission data, the temporal region between about 1 and

10 |ts is consistent with a proposed mechanism in which the particulate-phase is

preferentially enriched by the loss of gas-phase species, and is then additionally excited

(i.e., strong coupling) by the second laser pulse (Laser 2). At relatively shorter delay

times (< 500 ns), the dual-pulse enhancement is not as great because the rarefaction has

not yet developed, although plasma-laser coupling is still strong. At relatively longer

delay times (>10-20 [ts), the first plasma has undergone substantial decay, thereby

reducing the coupling of Laser 2, and the system is once again approaching conditions

corresponding to a single-pulse environment, as seen in the spectra of Figure 3-11.

3.1.4 Spectral Analysis of Borosilicate Glass Microspheres

To further examine the observed phenomena of dual-pulse enhancement with

aerosol analysis, in the context of the mechanisms offered above, additional experiments

were performed using significantly larger aerosol particles, namely the 2-[tm borosilicate

microspheres. This was done in an effort to determine if the plasma-particle interactions

with the dual-pulse configuration were further enhanced with increasing particle size.

Based on recent work with the identical particles, the overall time scale for particle

dissociation within the laser-induced plasma was estimated to be on the order of 15 tIS









[Carranza and Hahn 2002a], hence particulates are expected to be present during the

currently determined optimal dual-pulse laser-laser delays of between about 1 and 10 t.s.

For these experiments, suspensions of the borosilicate glass were nebulized as

described above, and single-shot spectra were analyzed for the presence of calcium

atomic emission using the same 393.4-nm Ca II atomic emission line. Measurements

were made using a single-pulse configuration (Laser 2 only) and a dual-pulse

configuration with a laser-laser delay of 250 ns. While this delay was not the optimal

value per the SNR experiments, the single-shot detection criteria makes use of the P/B

ratio [Carranza 2003]; hence the 250 ns was close to the optimal value per Figure 3-12,

and had the added advantage of slightly larger absolute signal counts to work with.

Because the particle hit rate was adjusted (i.e., aerosol number density reduced) to

be less than 100% to minimize multiple particle sampling with a single shot, any

ensemble-averaging would reduce the effect of dual-pulse enhancement on calcium

emission by averaging the calcium-containing spectra (i.e., particle hits) with non-

calcium containing spectra (i.e., particle free spectra ). Normally, this might be addressed

by using a conditional processing routine to separate out the spectra corresponding to

particle hits, and to then analyze only such spectra [Hahn 1997]. However, the easy

solution of identifying and ensemble-averaging the spectra of calcium-based particle hits

brings an additional problem. By enhancing the sensitivity of calcium detection with the

dual-pulse configuration, the particle hit rate is expected to increase. However, since

particle hits corresponding to the strongest calcium-emission signals are most likely to be

sampled with both single-pulse and dual-pulse configurations, the gains in sampling rate

with dual-pulse LIBS are expected to be made for the particle hits corresponding to the









weakest calcium emission signals. Therefore, comparing an ensemble-average of a larger

number of spectra that contain a larger percentage of weaker emission signals, to an

average containing a smaller number of spectra but with stronger emission signals is not a

valid comparison. Because development of a detailed algorithm to attempt to sort and

categorize spectra according to emission signal distributions was beyond the scope of the

present study, the borosilicate particle data were only analyzed in terms of the particle

sampling rate.

A comparison of the single-pulse to dual-pulse hit rate showed an increase by a

factor of 2.6 (250% increase) with the dual-pulse configuration. This was based on the

raw numbers of hits recorded using a threshold algorithm as previously described

[Carranza 2003]. In addition, all recorded single-shot spectra were then post-processed

using both calcium emission lines (393.4 and 396.9 nm) in an attempt to reject spectra

corresponding to false particle hits, as previously described [Hahn and Lunden 2000].

From the average spectra of all the hits, the average ratio of the peak-to-base for the

393.4 Ca II line to the 396.9 Ca II line was determined. All the hits that were inside the

range of the average ratio + a factor of 2 were decided as acceptable hits and all those out

of this range were rejected. Following this analysis, the ratio of particle hit rates with

dual-pulse LIBS as compared to the single-pulse (Laser 2 only) configuration was again

equal to 2.6, in exact agreement with the previous result. Overall, the borosilicate glass

particle experiments were consistent with the above fine-particulate experiments,

verifying that the dual-pulse LIBS configuration does produce an enhanced analyte

response for the micron-sized particles as well.









3.2 Particle Lens Study

An important topic when analyzing aerosols using LIBS is the introduction of the

analyte into the sample volume. Like in the first study, often a uniformly concentrated

aerosol and gas mixture flows through a large cross-sectioned tube in which the plasma

volume is situated in the center. However, if the cross-section were reduced in size, a

more concentrated flow could be achieved, and one would expect hit rates along with

analyte response to increase. This study focuses on the above comments, in which aerosol

response and hit rates were compared for sample injection using a flow lens to reduce the

flow cross-section down to 0.75 mm (.0295") diameter versus a diameter of 0.18" of a

standard 14" tube.

3.2.1 First Attempts

As mentioned in Chapter 2, experiments were first performed by drawing the

analyte through the two injection tubes at 1 1pm without a sheath flow, the following data

Table 3-1. Data representing the analyte response (Ca II) and the hit rates for the two
aerosol injection methods.
Steel Tube
Average P/B 1000 Shot Avg 1000 Shot Stdev Hit Average Hit Stdev
P/B 1 @ 393.4 nm 67.398 13.395 435.404 47.07
P/B 2 @ 396.9 nm 44.531 9.377 316.09 43.202

Hit # Average/1 000 shots Hit # Stdev/1000 shots
78.929 13.714

Particle Lens
Average P/B 1000 Shot Avg 1000 Shot Stdev Hit Average Hit Stdev
P/B 1 @ 393.4 nm 45.353 11.305 414.494 48.748
P/B 2 @ 396.9 nm 29.753 7.275 297.622 38.058

Hit # Average/i 000 shots Hit # Stdev/1000 shots
61.5 14.416









presented in this section describes these first attempts. Data, including hit rates and

analyte response for the 2-[lm borosilicate microspheres were gathered and compared for

the aerosol injection through the particle lens and the 14" thin walled tube. The first

ionized calcium (Ca II) atomic emission lines at 393.4 and 396.9 nm (0-25,414 cm-1 and

0-25,192 cm -, respectively) were used for all spectra analysis. Analyte response was

determined using the calculated peak-to-base of a 1000 shot averaged spectra and from

the averaged spectra of all the hits acquired in the 1000 shot set. Table 3-1 shows all the

averaged peak-to-base measurements and the average hit rates for both the particle lens

and 1/4 tubing. The number of hits as well as the analyte response were slightly better

when using the 1/4" tube rather than the particle lens. The average hit number per 1000

laser shots for the 1/4" tube was 78.9 (17.4% RSD) versus 61.5 (23.4% RSD) for the

particle lens. The 1000 shot average peak-to-base for the 393.4 Ca (II) line was 67.4

(19.8% RSD) and 45.4 (24.9% RSD) for the 1/4" tube and particle lens, respectively. Also

following a similar trend, the averaged P/B for each individual hit was 435.4 (10.8%

RSD) versus 414.5 (11.8% RSD) for the 1/4" tube and lens, respectively. The difference,

however, for all the mentioned results were not substantial and were rarely larger than

that of one standard deviation, therefore it was concluded that there was not any

advantage with either injection method.

Because such similar results were obtained, it was proposed that the particles were

entering the sample chamber and not exiting at the same rate, instead building up and re-

circulating back into the plasma volume. This was verified using a green diode laser to

perform light scattering to visualize the aerosol flow. Figure 3-15 [left], shows a straight

green line running across the sample chamber and in front of the particle lens. This line









was created by light scattering from the particles that were building up in the sample

chamber. This clearly shows that particles filled the entire chamber, making the lens

effects negligible. As mentioned in Chapter 2, to remedy this, an annular sheath flow was

situated around the particle lens and 14" tube to separate the particles exiting the tubes

with those re-circulating throughout the sample chamber. This can also be seen in Figure

3-15 [right]. The small point of light situated in the center of the tubes is light being

scattered by the particles exiting the particle lens, which are effectively isolated from

those that are re-circulating.














Figure 3-15. Images depicting the calcium aerosol light scattering of a green diode
laser. Looking down into the sample chamber, the beam path runs directly
across the center of the particle beam which is aligned in the center of the
12" annular tube. The green line represents the presence of particles. The
two images show the difference in particle location without [left] and with
[right] the annular sheath flow. As you can see, marked by the small spot
of scattered light at the center of the tubes, the image with the annular
sheath flow [right] effectively separated the particle exiting the lens with
those that were re-circulating in the chamber.

3.2.2 Sheath Flow Experimental Results

Like in the experiments without the sheath flow, particle hit rates, the peak-to-

base from the 1000 shot averaged spectra, and the average peak-to-base from each

individual hit of the 2-ptm borosilicate microspheres were used to quantify advantages or









disadvantages in using the particle lens. The same calcium lines, Ca II lines at 393.4 and

396.9 nm (0-25,414 cm-1 and 0-25,192 cm-1, respectively), were used to quantify the hit

rate and analyte response. It was believed that data from these experiments was more

conclusive since the effect of the re-circulating particles was removed.

A threshold conditional analysis approach was applied to obtain hit counts and

individual spectra when a hit occurred. This resulted in a number of hits for both the 14"

tube and the particle lens. To further eliminate any possible false hits, an extra filtering

process was used. From the average spectra of all the hits it was calculated that the

average ratio of the peak-to-base for the 393.4 Ca II line to the 396.9 Ca II line was

nominally 1.2 for both the 1" tube and the particle lens. An acceptable range for this

ratio was determined to be 1.2 + a factor of 2 (0.6-2.4). This process reduced the hit count

from 426 to 366 for the particle lens (a 14.1% reduction) and from 1769 to 1549 (a 12.4%

reduction). The similar reductions proved that false hits were just as likely to occur

whether a particle lens was used or not in s LIBS system.

Table 3-2 displays hit rates, averaged P/B of each spectra corresponding to a hit,

and the P/B of the averaged 1000 shot spectra for the 393.4 nm Ca II spectral line. The

data indicates no advantage when using a smaller particle injection orifice. Hits occurred

Table 3-2. Spectra and hit data comparing the /4" tube to the particle lens.
393.4 nm Ca II Line Data
Hit Average Hit Stdev 1000 Shot Average 1000 Shot Stdev
Hit Rate P/B P/B P/B P/B
1/4" Steel
Tube 0.0645 248.12 198.79 37.72 8.87
Particle Lens 0.0159 263.59 218.21 9.713 4.27

1.6 % of the time for the particle lens versus 6.4 % of the time for the /4" tube, a 75%

reduction. The 393.4 Ca II line peak-to-base of the 1000 shot average spectra was 9.7 for

the particle lens versus 37.7 for the /4" tube, also about a 75% reduction. This is expected







60


given the difference in hit rates between the two injection methods. Despite the decreased

hit rates and 1000 shot average spectra P/B, it was still hopeful that the spectra from a hit

would provide a better analyte response for the particle lens due to the fact that the

particle lens was more precise at delivering a particle into the core of the plasma, where

higher temperatures and electron densities exist. However, this was not the case, the

393.4 Ca II line average peak-to-base ratio for every individual hit were only slightly

better for the particle lens (263.6 with Stdev of 218.2) than with the 1/4" tube (248 with a

Stdev of 198.8), corresponding to only a 6% increase. With relative standard deviations

nearing 80%, the slight increase in the averaged hits P/B for the particle lens was not


400


350 --- ---1---------- -----
S 1/4" Tube
Se Particle Lens
300


250 --- ------------------ ----- -- -------


0

1 5 0 - - - -- - -
1 0 0 .: - - - - -- -: --: -:- -:- - :- -- - :- -- ---


100


50 -------


01
0 100 200 300 400 500 600 700 800 900 1000 1100
Peak-to-Base Range


Figure 3-16. Histogram categorizing the peak-to-base ratio of each individual hit for the
1/4" tube and the particle lens. The overall intensity is down for the particle
lens due to the smaller hit rates, but the distribution for both methods
follow very similar trends.







61


sufficient enough to rule that implementing the particle lens in combination with a

conditional analysis approach could produce better limits of detection. To further

investigate the quality of each individual hit for the particle lens and /4" tube, a histogram

and probability plot were constructed and can be see in Figures 3-16 and 3-17,

respectively.

The relative shape and distribution of the histograms for the 4" tube and particle

lens are very similar. This also negates the expectation that the particle lens would

provide a better quality hit ratio. If that were the case, one would expect to see the

distribution weighted more for the higher peak-to-base ranges. The probability plot seen

1200


1000 -
--no lens
-----lens |

800 -
i i "

S600 ------- --- -


0' 400 -- -- -----------------------
400*


200- ---------------------------I---------





200
-200 I-I-I- I-I I I I I-I I I I
.01 .1 1 5 10 2030 50 7080 9095 99 99.999.99
Percent


Figure 3-17. Probability plot comparing the distribution of the P/B ratios for all the
individual hits collected after the conditional analysis approaches were
performed.









in Figure 3-17 also clearly demonstrates this fact. Both the particle lens and /4" tube

follow the exact same distribution up to the 50% mark of a P/B equaling 200. After this,

the particle lens begins to slightly diverge, increasing a little faster than the /4" lens,

elucidating the slightly higher average P/B for the lens shown in Table 3-2. However, the

divergence of the lens is not enough to make the case that it can contribute to better

analyte response.

3.2.3 Summary and Particle Counting Results

In review, the particle lens produced much lower hit rates which correlated to the

much lower peak-to-base ratios of the averaged 1000 shot spectra. The average hits

spectra for both injection methods produced P/B ratios of very similar values. Each hit

was analyzed in anticipation that a higher percentage of the hits would contain larger P/B

ratios for the lens than the 1/" tube. This was not the case, which is evident in Figures 3-

13 and 3-14. Basically, when a hit occurred for either the lens or the 4" tubing (hits being

4 times more likely for the 14" tube) similar analyte response was found.

To clarify the above data, particle counting experiments were used to determine

whether the same amount of particles were exiting both the lens and the 14" tube. Pulling

1.0 1pm of the analyte through the lens and the tube, particles were counted using the

light scattering device. Considering all particles with diameters greater than 2 [tm, the

lens produced an average particle rate through the sample chamber of 154.16

particles/min with a standard deviation of 31.9 particles/min while the 4" tube produced

an average rate of 229.15 particles/min with a standard deviation of 57.3 particles/min.

This corresponded to the particle lens transmitting about 67% of the particles of the 4"

tube. This value of 67% transmission for the particle lens was almost identical with

preliminary model predictions done by Aerodyne Research, Inc. for 3 [im particles









flowing through the lens at 1 1pm [Wormhoudt 2006]. The transmission difference is a

large contributor for the discrepancy in the hit rates between the two injection methods

presented in Table 3-2, and could be attributed to particles sticking on the inner surface of

the particle lens when cross-sections were reduced by the smaller concentric tube.

Since 67% of entering particles were being transmitted through the focusing lens,

the question was, what else was contributing to the almost 75% reduction in hit rates

measured with the lens? Using a probability analysis and a Poisson distribution, while

assuming the flows had uniform particle distributions, predicted sampling rates were

determined for both the lens and the tube. Eq. 3.1 represents the Poison probability (Po)

that zero particles will be sampled. Subtracting that from one, as in Eq. 3.2, gives the

probability (Pn) that any number of particles will be sampled in the plasma, where '

represents the average number of particles per plasma volume and can be calculated, as in

Eq. 3.3, by multiplying the particle number density (N) by the effective plasma

Po=exp(-p) (3.1)

Pn=l-exp(-p) (3.2)

N=N*V (3.3)

volume (V).The above transmission values along with the known flow rate of 1 1pm were

used to determine the particle number density that was exiting both the lens and the tube.

For the lens Niens=0.15 particles/cc and for the tube Ntube=0.23 particles/cc. The

calculated number densities are not representative of the actual number densities due to a

number of reasons: (1) there was also a sheath flow diluting the actual number densities

that was not accounted for, (2) the particle counter had uncertainty, and (3) not all

particles exiting the lens and tube were going into the counter (the majority actually









entering the vacuum pump). However, when the two numbers are compared in a ratio, it

is significant and can be related to the ratio of the measured sampling rates in Table 3-2.

The effective statistical plasma diameter was previously determined as 1.3 mm [Carranza

and Hahn 2002b] which was used to calculate the effective plasma volume for the 1"

tube, Vtube 0.00123 cc. For the lens, the diameter of the flow exiting was smaller than the

effective plasma diameter of 1.3 mm, and therefore the effective plasma volume had to be

calculated as a cylindrical section of the plasma with a diameter equaling 0.75 mm

(diameter of the lens orifice) and a length of 1.3 mm (diameter of the plasma). The

calculated effective plasma volume for the lens was Viens=5.74E-4 cc. Figure 3-18

illustrates the reduction in the effective plasma volume when using the particle lens.

Plugging in the effective plasma volumes and number densities into Eqn. 3.3 to find i

- - -

Thin Walled \ Particle
1/4" Tube Focusing Lens




1 Plasma 0 0.75 mm
S01.3 mm Plasma

A B

Figure 3-18. The thin walled 14" tube [A] allowed for the maximum effective plasma
volume, represented by the shaded region, to excite the aerosol samples,
while the particle lens [B] reduced the effective plasma volume which
aided in reducing the particle hit rates.

and then into Eqn. 3.2, predicted sampling rates for each injection method were

calculated as Pnlens=0.89 particles/1000 shots and Pntube=2.8 particles/1000 shots for the

lens and /4" tube, respectively. This gave a ratio (Pnlens/Pntube) of about 0.31, which

closely compares to the measured ratio of 0.25. This means that the combination of the









reduced particle transmission with the reduction in the effective plasma sampling volume,

almost completely accounted for the large differences in particle sampling rates.

Now that the hit rate data has been explained, the question now becomes, why is

not the analyte response for the individual hits much larger for the lens than the 14" tube?

One reason could be explained through preliminary research done by Aerodyne in which,

measured particle beam profiles indicated that the particles had tendencies to disperse in

an annular direction [Wormhoudt 2006] therefore creating high concentrations of

particles possibly missing the hotter and more ionic core of the laser induced plasma,

rendering analyte responses similar to that of the 1/4" tube. Also, although the particle

streamline was set to go through the center of the plasma, particles could be sampled at

any point along the streamline path within the 1.3 mm plasma diameter. This meant that

the probability of sampling a particle in the center of the plasma along the streamline is

less than sampling a particle at the outer edge (before and after the center) of the plasma

along the streamline. More simply stated, even though the lens points the particles in the

direction of the center of the plasma, it is still unlikely to have a particle located directly

in the center when the plasma is formed.














CHAPTER 4
CONCLUSIONS AND PROPOSED FUTURE WORK

In this study two methods were investigated to assess their potential to increase the

analyte response and sampling rates for laser-induced breakdown spectroscopy, notably

for aerosol analysis. The two methods investigated were (1) applying a dual-pulse laser

configuration system, which has been proven to increase analyte response on liquid and

solid samples, for gaseous and aerosol analyte systems; and (2) using a particle lens to

focus aerosol particles directly into the laser-induced plasma in hopes that sampling rates

and analyte response would be enhanced. In summary, the two studies have accomplished

the following:

Dual-Pulse Study

Based on the current analysis, clearly the dual-pulse LIBS approach is applicable to
aerosol systems, namely the analysis of particulate-phase analytes. Under such
conditions, a 4-fold analyte signal (i.e., P/B and SNR) enhancement was achieved. It
is concluded that the system shows promise as a way to improve detection limits for
real-time aerosol sensing applications, which might justify the added system
complexity of dual-pulse configurations for critical sensing needs.

On the other hand, a rather poor signal response (both P/B and SNR) was realized
when applying dual-pulse LIBS for analysis of strictly gas-phase species. This result,
however, along with the transmission data, provides additional insight into the
physics of the plasma-analyte interactions, further supporting the concept of
preferential analyte depletion within the expanding plasma for pure gas-phase
analysis that has been explored in earlier works. Such an effect is the opposite of
solid-phase analytes, which are preferentially enriched, allowing for a dual-pulse
scheme to enhance the analyte signal (i.e., solid phase) to noise (i.e., gaseous phase)
ratio.

Particle Lens Study

The particle lens study revealed no advantage in using a more complex aerodynamic
lens tube over a traditional 1/4" thin walled tube. A conditional analysis approach was









used to determine sampling rates, which were reduced with the particle lens by almost
75%, subsequently reducing analyte response from a 1000 laser pulse averaged
spectra by almost 75%. In addition, each individual hit from the particle lens and the
/4" tube were examined and compared, revealing similar responses for both particle
injection methods. Therefore, the particles that were actually sampled by the plasma
produced similar signals.

Particle transmission along with the reduction in the effective plasma sampling
volume, due to the shrinking of the flow stream, have been considered as the major
contributors for the disparity in sampling rates. The particle lens transmitted about 2/3
of the amount of particles transmitted by the 14" tube alone, and reduced the effective
plasma volume by nearly 50%. Reasons for the similar analyte response of the
individual hits include possible flow alterations when exiting the lens orifice causing
the particle to miss the plasma core, and the fact that sampling a particle in the center
of the plasma even with a lens in place is still a low probability event.

To further understand the physics behind the data acquired, additional work must

be carried out for both studies. Proposed experiments and examination include:

Dual-Pulse Study

Running similar procedures as in this study, but with particle analytes with varying
characteristics (i.e., size, melting point, etc.) to more completely understand the
plasma-particle interactions that occur and to determine the circumstances when a
dual-pulse system could be optimized to achieve the best detection limits.

Further experiments to elucidate the preferential accumulation of solid particles inside
the laser-induced plasma should be carried out. This could include plasma imaging
with multiple cameras so that a series of images could determine the particles effect
with the expanding plasma, as well as performing light scattering experiments to
more clearly understand the physics behind the achieved signal enhancement for the
aerosol based analyte.

Particle Lens Study

Flow simulation could be run to determine the reason for the reduced particle
transmission through the particle lens. Also, a detailed flow simulation could be done
at the exit orifice of the particle focusing lens to determine if there are secondary flow
effects causing the particles to disperse radially rather than in a streamline.

Running experiments and calculations to optimize the particle lens design so that the
cross-section of the particle flow exiting the lens will match with the maximum
plasma volume.
















APPENDIX A
COMPONENTS OF SPECTROSCOPIC SYSTEMS

Table A-1. List of components used in the dual-pulse study.
Device Manufacturer Description
Equipment
Nd:YAG laser Big Sky Laser 1064 nm, 10 ns pulse, 5 Hz, 290
Technologies mJ/pulse
Nd:YAG laser Continuum 1064 nm, 5 Hz, 100 mJ/pulse
Spectrometer Acton Research 0.275 m spectrometer, 2400
Corporation grooves/mm, 195 nm-2800 nm
iCCD Princeton Instruments Intensified CCD, 200 row chip
iCCD Chiller Refrigerator
re-circulator
Software Custom Metal Emissions Labview
Program
Delay Pulse Generator Stanford Research 4 channel digital delay/pulse
Systems, Inc. generator
Fiber Optic Acton Research 6', high optical grade, 17 fiber
Corporation bundle, 1.5 mm diameter
Optics
Laser 1 Optical Telescope CVI Laser Corporation x2.5, two 1064 UV grade AR
lenses
Laser 2 Optical Telescope CVI Laser Corporation xl.7, two 1064 UV grade AR
lenses
Elliptical Pierced Mirror CVI Laser Corporation UV-grade AR enhanced
Laser 1 Focusing Lens CVI Laser Corporation 100-mm focal length, 50-mm
dia, UV-grade, 1064 nm AR
Laser 2 Focusing Lens CVI Laser Corporation 75-mm focal length, 50-mm dia,
UV-grade, 1064 nm AR
1064 nm dichroic Mirror CVI Laser Corporation 45 degrees, 1064 dichroic
mirror, 2" diameter
Aperture Newport 1 aperture
Quartz Window Huntington UV-grade quartz window
Collection Lens CVI Laser Corporation 75-mm focal length, 50-mm dia,
UV-grade, 1064 nm AR









Table A-2. List of components used in the particle lens study.
Device Manufacturer Description
Equipment
1064 nm frequency Continuum Surelite II 532/355 nm, 5 Hz, 60 mJ/pulse
doubled/tripled Nd:YAG
Laser
Spectrometer Acton Research 0.275 m spectrometer, 2400
Corporation grooves/mm, 195 nm-2800 nm
iCCD Princeton Instruments Intensified CCD, 200 row chip
iCCD Chiller Refrigerator re-
circulator
Software Custom Metal Emissions Labview
Program
Vacuum Pump Thomas Compressors
and Vacuum Pumps
Particle Counter Particle Counting Light scattering particle
Systems counter, 0.2 1pm
Particle Lens Aerodyne research, Inc 60 cm long, 0.75 mm exit
orifice
Fiber Optic Acton Research 6', high optical grade, 17 fiber
Corporation bundle, 1.5 mm diameter
Optics
355 nm dichroic mirror CVI Laser Corporation 45 degree, 355 nm dichroic
mirror, 2" diameter
Aperture Newport 1 aperture
Focusing Lens CVI Laser Corporation Plano-convex, 100 mm UV-
grade, 2" diameter
Collection Lens CVI Laser Corporation Plano-convex, 100 mm UV-
grade, 2" diameter
Square Pierced Mirror Rolyn Optics 100 mmx100 mmx mm center
pierced-0.5"+0.2" diameter
Quartz Window Huntington UV-grade quartz window















APPENDIX B
PARTICLE LENS CALCULATIONS


Table B-1. Results of calculations used to determine predicted sampling rates for the
particle lens and /4" steel tube.

Tube Particle Particle
Diameter Tube Area Volume Flow Rate Transmission Velocity

(m) (mA2) (mA3/s) (particles/s) (m/s)

Particle
Lens 7.50E-04 4.42E-07 1.67E-05 2.569 37.73

1/4"
Tube 4.57E-03 1.64E-05 1.67E-05 3.7616 1.0152





Effective
Particle Plasma Pn (calculated Measured
Density (N) Vol. it hit rate) Sampling rate

(particles/plasma (particles/1000 (particles/1000
(particles/cc) (cc) volume) shots) shots)
Particle
Lens 0.1546 5.74E-04 8.88E-05 8.88E-02 1.59E-02

1/4"
Tube 0.229 0.00123 2.82E-04 2.82E-01 0.064542

Ratio 31.49% 24.66%

















LIST OF REFERENCES

Angel S.M., Stratis D.N., Eland K.L., Lai T., Berg M.A., Gold D.M., Fresenius' Journal
Analytical Chemistry 369 (2001) 320-327.

Bilajic D., Corsi M., Cristoforetti G., Legnaioli S., Palleschi V., Salvetti A., and Tognoni
E., Spectrochimica Acta Part B, (2002) 339-353.

Buckley, S.G.; Johnsen, H.A.; Hencken, K.R.; Hahn, D.W., Waste Manage. 20 (2000)
455-462.

Carranza J.E., Fisher B.T., Yoder G.D., and Hahn D.W., Spectrochimica Acta Part B, 56
(2001) 851-864.

Carranza J.E., Hahn D.W., Spectrochimica Acta Part B, 57 (2002a) 779-790.

Carranza J.E., Hahn D.W., J. Anal. Atomic Spectroscopy 17 (2002b) 1534-1539.

Carranza J.E., Hahn D.W., Analytical Chemistry 74 (2002c) 5450-5454.

Carranza J.E., lida K. and Hahn D.W., Applied Optics, 42 (2003) 6022-6028.

Cheng M-. D., Fuel Processing Technology, 65 (2000) 219-229.

Ciucci A., Corsi M., Palleschi V., Rastelli S., Salvetti A., and Tognoni E., Applied
Spectroscopy 53 (1999) 960-964.

Coloa F., Lazic V., Fantoni R., Pershin S., Spectrochimica Acta Part B: Atomic
Spectroscopy 57 (2002) 1167-1179.

Corsi M., Cristoforetti G., Giuffrida M., Hildago M., Legnaioli S., Palleschi V., Salvetti
A., Tognoni E., Vallebona C., Spectrochimica Acta Part B: Atomic Spectroscopy
59 (2004) 723-735.

Cremers D. A. and Radziemski L. J., Applied Spectroscopy, 39 (1985) 57-63.

Dudragne L., Adam Ph., Amouroux J., Applied Spectroscopy, 52 (1998) 1321-1327.

Essien M., Radziemski L.J., and Sneddon J., Journal of Analytical Atomic Spectrometry,
3 (1988) 985-988.

Ferioli F., Puzinauskas P.V., Buckley S.G., Applied Spectroscopy 57 (2003) 1183-1189.









Fisher B.T., Johnsen H.A., Buckley S.G., and Hahn D.W., Applied Spectroscopy, 55
(2001)1312-1319.

Gautier C., Fichet P., Menut D., Lacour J.-L., L'Hermite D., Dubessy J., Spectrochimica
Acta Part B: Atomic Spectroscopy 60 (2005) 265-276.

Gleason R.L. and Hahn D.W., Spectrochimica Acta Part B 56 (2001) 419-430.

Gornushkin I.B., Kazakov A.Y., Omenetto N., Smith B.W.,Winefordner J.D.,
Spectrochimica Acta Part B 59 (2004) 401-418.

H. Kaiser, Spectrochimica Acta, 33B (1978) 551.

Hahn D. W., Flower W. L., and Hencken K. R., Applied Spectroscopy, 51 (1997) 1836-
1844.

Hahn D.W., Applied Physics Letters, 72 (1998) 2960-2962.

Hahn D.W. and Lunden M.M., Aerosol Science and Technology, 33 (2000) 30-48.

Hahn D.W., Carranza J.E., Arsenault G.R., Johnsen H.A., and Hencken K.R., Review of
Scientific Instruments, 72 (2001) 3706-3713.

Hohreiter V., Ball A., Hahn D.W., J. Analytical Atomic Spectroscopy, 19 (2004a) 1289-
1294.

Hohreiter V., Carranza J. E., Hahn D. W., Spectrochimica Acta Part B: Atomic
Spectroscopy 59 (2004b) 327-333.

Hohreiter V. and Hahn D. W., Analytical Chemistry, 77 (2005a) 1118-1124.

Hohreiter V., Hahn D.W., Spectrochimica Acta Part B: Atomic Spectroscopy 60 (2005b)
968-974.

Hohreiter V. and Hahn D. W., Analytical Chemistry, 78 (2006) 1509-1514.

Ingle J. D. J., and Crouch S. R. SpectrochemicalAnalysis, (1988) Prentice Hall,
Englewood Cliffs, NJ.

Ismail M. A., Imam H., Elhassan A., Youniss W. T., and Harith M. A., J. Analytical
Atomic Spectroscopy, 19 (2004) 489.

Kaiser H., Spectrochimica Acta, 33B (1978) 551.

Lee W-. B., Wu J., Lee Y-. I., and Sneddon J., Applied Spectroscopy Reviews, 39 (2004)
27-97.

Lencioni D.E., Applied Physics Letters, 23(1) (1972) 12-14.









Lindner H., Koch J., Niemax K., Analytical Chemistry, 77 (2005) 7528-7533.

Martin M. Z. and Cheng M-D., Applied Spectroscopy, 54 (2000) 1279-1285.

Neuhauser R. E., Panne U., Niessner R., Petrucci G. A., Cavalli P., Omenetto N.,
Analytica Chimica Acta. 346 (1997) 37-48.

Neuhauser R.E., Panne U., Niessner R., and Wilbring P., Fresenius Journal of Analytical
Chemistry, 364 (1999) 720-726.

Nunez M.H., Cavalli P., Petrucci G., and Omenetto N., Applied Spectroscopy. 54 (2000)
1805-1816.

Peng L. W., Flower W. L., Hencken K. R., Johnsen H. A., Renzi R. F., and French N. B.,
Process and Control Quality. 7 (1995) 39-49.

Radziemski L.J. and Loree T.R., Plasma Chemistry and Plasma Processing, 1 (1981)
281-293.

Radziemski L.J., Loree T.R., Cremers D.A., and Hoffman N.M., Analytical Chemistry,
55 (1983) 1246-1252.

Radziemski L. J., Microchemical Journal, 50 (1994) 218-234.

Radziemski L.J, Spectrochimica Acta Part B, 57 (2002) 1109-1113.

Sattman R., Sturm V., Noll R., Laser, J. Phys., D. Appl. Phys. 28 (1995) 2181-2187.

Scaffidi J., Pearman W., Lawrence M., Carter J., Colston B.W., Angel S.M., Applied
Optics, 43 (27) (2004) 5243-5250.

Schechter I., Analytical Science and Technology, 8 (1995) 779-786.

Schechter I., Reviews in Analytical Chemistry, 16 (1997) 173-298.

Singh J.P., Yueh F.Y., Zhang H.S., and Cook R.L., Process and Control Quality, 10
(1997) 247-258.

Smith D.C., Journal of Applied Physics, 48 (1997) 2217-2225.

Smith B.W., Hahn D.W., Gibb E., Gornushkin I., and Winefordner J.D., KONA Powder
and Particle No 19 (2001) 25-33.

Sneddon J. and Lee Y-I, Analytical Letters, 32 (1999) 2143-2162.

St-Onge L., Sabsabi M., Cielo P., Spectrochimica Acta Part B: Atomic Spectroscopy, 53
(1998) 407-415.

Stratis D.N., Eland K.L., Angel S.M., Applied Spectroscopy 54 (9) (2000) 1270-1274.









Stratis D.N., Eland K.L., Angel S.M., Applied Spectroscopy 55 (2001) 1297-1303.

Tognoni E., Palleschi V., Corsi M., Cristoforetti G., Spectrochimica Acta Part B, 57
(2002)1115-1130.

Tran M., Smith B.W., Hahn D.W., Winefordner J.D., Applied Spectroscopy, 55,
(2001)1455-1461.

U. S. Environmental Protection Agency, Federal Registration 61, 77, (1996) 17357-17358.

Uebbing J., Brust J., Sdorra W., Leis F., Niemax K., Applied Spectroscopy, 45 (9) (1991)
1419-1423.

Weyl G.M. Physics of Laser-Induced Breakdown: An Update, in Laser-InducedPlasmas
and Applications, edited by L.J. Radziemski and D.A. Cremers (1989), Marcel
Dekker, New York, Chapter 1.

Winefordner J. D., Gornushkin I. B., Correll T., Gibb E., Smith B. W., and Omenetto N.,
Journal of Analytical Atomic Spectrometry, (2004), 19, 1061.

Wormhoudt J. C., Particle Lens Guide. Boston, MA: Aerodyne Research, Inc. (2006)

Zhang H., Yueh F.Y., and Singh J.P., Applied Optics, 38 (1999) 1459-1466.















BIOGRAPHICAL SKETCH

Bret Windom earned his undergraduate degree in Mechanical Engineering at the

University of Florida in 2004. The work presented here is a culmination of his Master's

studies, while he is currently pursuing his doctoral degree under Dr. David Hahn at the

University of Florida.