UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An attampt to measure the electron temperature of a Xenon Chloride excimer laser discharge plasma by… Hughes, Michael K. Y. 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_hughes_michael.pdf [ 5.04MB ]
JSON: 831-1.0085616.json
JSON-LD: 831-1.0085616-ld.json
RDF/XML (Pretty): 831-1.0085616-rdf.xml
RDF/JSON: 831-1.0085616-rdf.json
Turtle: 831-1.0085616-turtle.txt
N-Triples: 831-1.0085616-rdf-ntriples.txt
Original Record: 831-1.0085616-source.json
Full Text

Full Text

AN ATTEMPT TO MEASURE THE ELECTRON TEMPERATURE OF A XENONCHLORIDE EXCIMER LASER DISCHARGE PLASMA BY THOMSONSCATTERINGbyMICHAEL KON YEW HUGHESB.Sc. (Eng), Queen's University (Kingston), 1990A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinTHE FACULTY OF GRADUATE STUDIESDepartment of PhysicsWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAJuly 1993© Michael Kon Yew Hughes, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.Department of PAySiCSThe University of British ColumbiaVancouver, CanadaDate 12 41-e/ / DE-6 (2/88)AbstractAn electronic timing system was developed so that a commercial Nd:YAGregenerative amplifier could be triggered remotely with a BNC cable and pushbuttonswitch. This system was used to accurately time a Xenon Chloride transverse dischargelaser to the short, high power Nd:YAG pulse to within 5 ns.Attempts were made to determine the electron velocity distribution of the dischargeplasma of this Xenon Chloride laser using the method of Thomson scattering. TheThomson scattering parameter a was 0.22. A short, 532 nm pulse was injected into theexcimer laser and the backscattered light was analysed. The scattered beam was imagedonto the entrance slit of a spectrometer and the spectrally dispersed output was collectedby either a streak camera or a photomultiplier array.The injection laser pulses were produced by amplifying the 100 Ps pulses from amodelocked Nd:YAG laser in a regenerative amplifier giving probe pulses ofapproximately 40 mJ in 100 Ps at the excimer laser.Aside from the signal intensity at shifted wavelengths which was measured to give anestimate of the electron temperature, the relative timing of the probe pulse to the excimercurrent was measured to give a time profile of the temperature. The injection pulse powerand excimer laser current were also needed to ensure that results behaved as predicted inrelation to these parameters.The temperature measurements proved to be impossible due to extremely high noiselevels, a lack of resolution of the spectrometer and possibly laser heating effects.11Table of Contents:Abstract^ iiTable of Contents^ iiiList of Figures vList of Tables^ vii1 Introduction 1^1.1 Overview of Laser Operation^ 31.2.1 Laser Resonators 41.2.2 Mode Locking 51.2.3 Acousto-optic Modulation^ 61.2.4 Pockels Cells^ 71.2.5 Second Harmonic Generation 81.2 Historical Overview 101.3 XeC1Laser Chemistry and Kinetics^ 112 Theory^ 152.1 Definition of Electron Temperature 152.2 Thomson Scattering^ 153 Apparatus and Setup 203.1 Overview^ 203.2 Excimer Laser 203.2.1 Excimer Laser Body Construction^ 203.2.2 Excimer Laser Circuit^ 233.2.3 Excimer Laser Operation ,^253.3 Injection Pulse Production 263.3.1 YAG Oscillator^ 263.3.2 Regenerative Amplifier 273.3.3 Steering Optics 293.4 Collection Optics^ 293.5 Optical System Alignment 313.6 Scattered Light Dispersion^ 333.7 Streak Camera 5^343.8 Photomultipliers^ 373.9 Electronic Measurements^ 404 Electronic Timing and Synchronization 434.1 Regenerative Amplifier 434.1.1 Control Unit^ 44iiiTable of Contents4.1.2 Power and Capacitor Banks^ 464.1.3 Laser Bench Controls 464.2 Delay Box^ 474.3 Streak Camera 474.4 Measurement Apparatus^ 485 Measurements and Results 495.1 Streak Camera^ 495.2 Photomultiplier Tubes 535.3 Discussion 555.3.1 Resolution^ 555.3.2 Noise 585.3.3 Laser Heating Effects^ 585.3.4 Conclusions^ 596 Future Work^ 60References 63ivList of Figures:1.1 Energy Diagram for a Four Level Laser^ 41.2 Bragg Scattering^ 61.3 Second Harmonic Generation Index Matching: Indices of Refraction for aNegative Uniaxial Crystal^ 91.4 Pathways for the formation of XeC1*^ 122.1 Scattering Definitions^  163.1 Experimental Setup 213.2 Cross Section of Laser Body^ 223.3 Simplified Laser Discharge Circuit 233.4 Laser Circuit^ 243.5 Diagram of the Regenerative Amplifier^ 273.6 Telescope Arrangement for Injected Pulse 293.7 Optical Alignment^ 313.8 Collection Lens Focusing Arrangement^ 323.9 Ebert Arrangement Spectrometer 333.10 Grating Definitions^ 333.11 Principle of Streak Camera Operation^ 353.12 Magnifying Arrangement for Dispersed Signal^ 363.13 Photomultiplier Tube Coupling with Fiber Optic Bundle^ 383.14 Photomultiplier Tube Spectral Dispersion Calibration Arrangement^ 393.15 Calibration of Photomultipliers^ 393.16 Photomultiplier Voltage for a Single Photon^ 403.17 Measurement Apparatus^ 41VList of Figures3.18 Rogowski Coil Current as Seen on Oscilloscope^ 424.1 "Single Shot" Mode^ 444.2 "Fixed" Mode 455.1 Wavelength Spectrum of Scattered Light^ 495.2 Streak Camera Image^ 505.3 Analysed Streak Camera Image^ 525.4 Background and Signal Time Signals to show Rayleigh Scattering^ 535.4 Typical Photomultiplier Tube Signals with an Excimer Discharge 545.5 Typical Photomultiplier Tube Signals Without Excimer Discharge^ 55viList of Tables:4.1 External Connector Pin Assignments for "Single Shot" Mode^ 46vii1 IntroductionThis work follows the acquisition of a regenerative amplifier which made possible theproduction of short, high power, 1064 and 532 nm light pulses at rates up to 10 Hz. Touse this new amplifier system, it was necessary to understand its internal timing and todetermine the synchronization pulses available to external timing devices. This knowledgewas used to accurately synchronize the regenerative amplifier, the discharge laser pulses,and the measurement devices.In previous experiments, synchronization was left to chance. This was possiblebecause probe pulses were produced at intervals of tens of nanoseconds making it possibleto capture a pulse within the discharge laser pulse which was tens to hundreds ofnanoseconds long. Such synchronization is not possible with a repetition rate of only10 Hz. This thesis describes how such timing was achieved.This work has applications in many experiments in which accurate synchronization isnecessary. It is relevant to experiments investigating laser produced plasmas or work inwhich the pulses of two lasers must be synchronized such as in the Thomson scatteringexperiment to be described. A method for accurate timing is needed to fully utilize thehigh power short pulses now available.Once the system was in place to limit the timing uncertainty to under 5 ns, attemptswere made to use the method of Thomson scattering to investigate a Xenon Chlorideexcimer laser. More specifically, the objective was to determine the electron velocitydistribution of the discharge plasma as a function of time. This work follows upon thework of Elezzabil who constructed the laser and studied its electron density and dischargedynamics. The electron velocity distribution is essential for comparing experimental datato theoretical modeling. Most theoretical studies have predicted a significant deviationfrom Maxwellian electron velocity distributions which is not found in most experimental1Chapter 1: Introductionresults. This discrepancy must be explained through more theoretical and experimentalwork to understand excimer laser kinetics.To perform this experiment a short, 532 nm pulse was injected into the excimer laserand the backscattered light was analysed. The scattered beam was imaged onto theentrance slit of a spectrometer and the spectrally dispersed output was collected by eithera streak camera or a photomultiplier array.To estimate the electron temperature, it was necessary to analyse the distribution ofsignal intensities at shifted wavelengths. The relative timing of the probe pulse to theexcimer current was needed to give a time profile of this temperature. The injection pulsepower and excimer laser current were also needed to ensure that the measurement datacould be normalized and therefore compared. While many attempts were made, electrontemperature measurements proved to be impossible. Extremely high noise levels, anunexplained lack of resolution of the spectrometer, and possibly laser heating effects madethe measurements difficult.The only known studies of an excimer laser using Thomson scattering have been doneby a Japanese group in 19892 and 1991.3 The earlier experiment measured the density andelectron velocity distribution for Kr/Ne, Kr/He, and Kr/Ne/HC1 mixtures using a 0.85 J,30 ns, ruby laser probe beam. These results showed a higher electron temperature for neonthan for helium based mixtures. This group's later work used a frequency doubled, 0.7 J,10 ns, Nd:YAG laser to make measurements on the same laser studied previously using a902 scattering angle. This time measurements were also made on mixtures containingxenon; however, the xenon measurements suffered from a laser heating effect which madeany results invalid. Nonetheless, for the other mixtures, an essentially Maxwellian velocitydistribution was found, in disagreement with what was predicted.The remainder of this chapter contains a brief overview of laser operation, adiscussion of previous work done in this field, and a discussion of the operation of a XeC1discharge laser. The following chapters discuss, in turn, the theory of Thomson scattering,2Chapter 1: Introductionthe apparatus, the synchronization system, the measurements and results, and finally, anoutline of the work that would be necessary to make this experiment successful.1.1 Overview of Laser OperationIn this section, a brief overview of laser physics will be given since the rest of thiswork requires such knowledge. The principles of laser physics is more thoroughly coveredin texts such as Millonni and Eberly's Lasers4 , or Yariv's Quantum Electronics5As is well known, the term "laser" is an acronym for Light Amplification by theStimulated Emission of Radiation. For the purposes of this experiment, the most usefulproperties of a laser are the possibilities of a narrow wavelength spectrum, high intensity,and very short pulse length.In a simplified treatment, we can consider a medium in which electronic transitionsare possible between an excited state centred around an energy E2 and a lower statecentred around an energy El. The upper level has a population N2 and quantumdegeneracy g2 the lower level has a population N1 and quantum degeneracy gl.Interactions are possible with Np photons of energy E2-E1. Photons can be absorbed byNiN„the lower level with a probability proportional to ^ ; photons can be spontaneouslyg1N2created with a probability proportional to —; or photons can be emitted with the sameN2Npphase and direction as existing photons with a probability proportional to ^ . Thisg2last process is known as stimulated emission. The probability for a photon at the transitionfrequency to stimulate emission or to be absorbed by the lower level is proportional to thenumber of atoms in either state.N2of incident radiation, — mustg2This implies that in order for there to be any amplificationNibe greater than —; this condition is known as inversion.gFor systems in thermal equilibrium, inversions do not occur naturally as the population ofany given level is given by the Boltzmann factor: exp(-E/kbT). States with higher energies3hv--a.4 IFigure (1.1) EnergyDiagram for a FourLevel Laser.Chapter 1: Introductionhave less chance of being occupied. By various means inversionscan be created by pumping higher level states.Some methods to pump a medium are by focusing theradiation from flash lamps into the gain medium causing opticalabsorption, by passing an electrical current through the mediumby transferring energy through collisions with electrons, or bycombining certain chemical species which form an excited state.The rate of pumping must be sufficient to make up for the loss of excited states due tostimulated or spontaneous emission or decay.Most practical laser mediums have more than just two levels. One of the systems weare dealing with is the four level Nd3±:YAG system (see figure (1.1)). In a four levelsystem, inversion is easily achieved if the spontaneous decay probability from 2 to 1 isgreater than the probability from 3 to 2 so that the upper level will tend to be morepopulated in comparison when the excited atoms spontaneously cascade in energy fromlevel 4 to 3 to 2 to 1. Also, pumping is made easier if the 4 to 3 transition is fast so thatintense pumping doesn't de-excite the upper laser level. The excimer laser system will bedescribed in more detail in the next section.1.2.1 Laser ResonatorsA resonator surrounding the gain medium is necessary to provide high intensities, toselect specific modes, and to produce a narrow beam of laser light. A resonator can beformed by placing end mirrors around the gain medium. Feedback also ensures that theintensity and phase of the laser output are fairly consistent. Usually for a laser cavity, oneof the mirrors is fully reflective and the other is partially reflective and lets the laser lightout of the cavity. For most applications, a "stable" optical resonator is desirable. Stabilityimplies that a group of photons reflecting between the end mirrors will repetitively pass4Chapter 1: Introductionthrough the gain medium and will not be lost in the radial direction. One method toachieve this requires that at least one of the mirrors must be concave and that the distancebetween the mirrors must be less than the sum of the radii of curvature. For gain to occur,any losses through the mirrors or through other sources must be made up by the gain inthe gain medium.Any modes which do not match the resonance of the cavity will not be amplified.1.2.2 Mode LockingShort pulse production is necessary for our experiment: both to provide good timeresolution as well as to efficiently produce high intensity pulses. Mode locking is used tomake short pulses in the resonator. As the name suggests, mode locking refers to thetechnique of forcing all the longitudinal modes in a laser cavity to oscillate in phase. Howthis generates short pulses can easily be seen if we represent the electric field of eachlongitudinal mode in the laser simply asEn = Eo expfi[con (t —^+ (pn 1.P-1If the total electric field, E Total, is E Total =^En we find that for a large number ofn=—(p-1)modes of comparable strength, we need zero phase difference between the modes forthere to be a pulse. We find for no phase difference between modes,exp[ipAco(t —1]'Total = EOand ETotai = 0 otherwise. Here Aw = S- and L is the cavity length. It can be shown thatthe shortest pulse length possible is approximately equal to 4L/pc; this is known as theFourier limited pulse. In practice, it is usually not possible to get such short pulses fromexp [iAco(t —^— 15 kskskDeflectedFigure (1.2) Bragg ScatteringChapter 1: Introductionmode locking and a technique called pulse compression is necessary to get Fourier limitedpulses.The most common way of mode locking a laser is to introduce optical losses to allmodes that are not in the desired phase. The modes that have the highest round trip gainwill dominate over all the others. Losses can be introduced by methods such as acousto-optic modulation.1.2.3 Acousto-optic ModulationAcousto-optic modulation is achieved by placing a crystal which is very efficient atpromoting coupling between photons and phonons inside the optical cavity and attachingit to a piezo-electric crystal. The frequency of the driving voltage is carefully adjusted sothat an acoustic standing wave is set up within the crystal. Photons passing through thecavity (with wavevector k) have a high chance of being deflected through absorption of aphonon (with wavevector k) so thatk±k, =kDeflected as shown in figure (1.2).There will be a significant proportion of lightdeflected for any significant density of phononsso only the pulse traveling in resonance with thezero amplitude crossings of the acoustic wave will be amplified. Another way we can lookat acousto-optic scattering is by assuming that the standing sound wave sets up a periodicvariation in the refractive index. This can be described asAn(x,t) 0= sin (ksx)sin (kst). Theincident light will then diffract from the variations in much the same way X-rays diffractfrom atomic planes in crystals. This is commonly known as Bragg scattering.6Chapter 1: Introduction1.2.4 Pockels CellsAnother important component in our optical system is the Pockels cell. This elementallows dynamic polarization control of a polarized light beam through the use ofbirefringent materials and the electro-optic effect. Birefringent materials are those whichexhibit different indices of refraction for different polarizations. The electro-optic effect isthe variation of the degree of birefringence with an electric field.In birefringent materials, there is a direction of propagation in crystals for which theindex of refraction does not depend on the polarization known as the optic axis (OA). Awave having a polarization perpendicular to this axis will have a constant index ofrefraction; this is known as the ordinary wave. A wave with the orthogonal polarizationand same wavevector is known as the extraordinary wave and its index of refractiondepends on its direction of propagation. If we think of a light wave as being composed oftwo separate components of orthogonal polarizations, one polarized in the ordinarydirection and the other in the extraordinary direction, the components will travel throughthe crystal with different velocities (due to different indices of refraction). We can see howbirefringent crystals can change polarizations if we look at a linearly polarized wave,E(z,t)=E 0 +E,, oriented so thatEo(z =0,0= 1E sin cotand^E„ (z = 0,t)=1Esin wt.After traveling through the crystal a distance z1,Eo(zi,t)=IEsin(cot—kozi)and^E„(zi,t)=)Esin(c)t—k„zi).If (k„ — k0)z1= 12 then the total field is circularly polarized and if (k„—k0)z1= it thenthe field is polarized in the orthogonal linear direction.Combining this effect with the electro-optic effect makes dynamic control over thepolarization possible. A Pockels cell contains an electro-optic crystal oriented in the7Chapter 1: Introductionproper direction and of the proper length so that when a known voltage is provided, thedesired change in polarization can be achieved. If we combine the Pockels cell with apolarizer, it is obvious that the combination can act as a switch for polarized light.1.2.5 Second Harmonic GenerationOne more component essential for our experiment is a second harmonic generator.The production of light at half the input wavelength is dependent on a non-linear electronresponse to an optic field. Usually, electrons in bound states are modeled to first order asbeing in a quadratic potential well: V = VID + ar2 . However, for the second order modelingof the anharmonic potential well, we need one more term in the potential:V = Vo + ar2 + br3 . Following Yariv, we can model the electric field forced motion ofelectrons in such a well as/ = eE0 (eicor ± e-icor)F+yr+co(2)r+Dr-^ ...(1.1)2mwith solutionr=1(qieicot+ 4,2e i2co t ± c. c.)2egp^1 where q1 = 2ril (00 — (02 — icoy—De2 El,^1and q2 = 2m2^ [0)12), _ 0)2 _ iwyr (w — 4w—5 (n — /any)...(1.2)y being the damping term. Therefore, it can be seen from equation (1.2) that including asecond order in the approximation of the electronic potential well leads to secondharmonic generation of a driving optical electric field.The efficiency, isHG, of second harmonic generation is given to first order bySM. 2 (A1(1-)2 ) 11SHG =P2co =...( 1.3)Po)( AkL 28OAFigure (1.3) SecondHarmonic GenerationIndex Matching:Indices of Refractionfor a NegativeUniaxial CrystalChapter 1: Introductionwhere Ak = k2w — 2e . For maximum efficiency, we want thisphase difference to approach zero. Intuitively, we can see whythis is necessary. Knowing that the second harmonic wave isgenerated in phase with the pumping wave, if there is no phasematching then the second harmonic generated at one point willinterfere with the second harmonic generated at all other points.The most common method for phase matching, in appropriatecrystals, is to send in the pump wave polarized on the ordinaryaxis and generate the second harmonic polarized on theextraordinary axis. The angle of the direction of propagation withrespect to the optic axis has to be carefully chosen so that thephase matching condition is fulfilled. Figure (1.3) illustrates this condition for a negativeuniaxial crystal (a crystal that has only one optic axis and where no < no). This conditioncan be described mathematically as-2^-2sin 2 Om = n (w) — n 0 (2 w)Small deviations from the phase matching condition can lead to dramatic losses in theconversion efficiency. So precise control of the orientation is necessary.The efficiency implied by equation (1.3) is in fact not achievable. The theory used toderive this equation assumes that the intensity of the beam at w does not change withdistance; but energy must of course be taken from the pump beam and put into the secondharmonic. The theoretical maximum conversion efficiency does indeed approach 100%with very high intensities but the damage threshold of real crystals, is below the necessaryintensities.—{n 0 (w)-2 — n e (2 ())-2 •9Chapter 1: Introduction1.2 Historical OverviewThe electron temperature of a plasma can be determined by various spectroscopicmethods6 or by Thomson scattering. Thomson scattering, the scattering of radiation fromfree electrons, first was used to study the ionosphere using radio waves. The advent oflasers in the early 60's provided the necessary powerful, monochromatic light sourcesneeded to perform such scattering in a laboratory. The very first studies were done byFiocco and Thompson;7 they studied light scattering from an electron beam. This workwas quickly followed by many studies of laboratory plasmas: for example see Thompsonand Fiocco8.The first excimer lasers were developed around 1972. Excimer lasers are of interest toscience and industry because of their high gain, high efficiency, and high power atultraviolet wavelengths. They have high gains since the lasing transition has a thermallyunstable ground state. This means that there is little absorption at the lasing frequencybecause there are very few molecules in the lower laser level. Many excimer laser systemshave been found; they can be divided into classes such as rare gas, rare gas halogen, metalvapor, or triatomic rare gas excimers9. The type we are interested in are the rare gas halideexcimers.Quickly following the announcements of the potential of rare gas monohalide systemsas laser medial°,11, the first such lasers appeared in 1975 12,13,14. They used electron beamsto produce the necessary excitation. To make electrical discharge pumping feasible, firstthe obstacle of inherent discharge instability in excimer systems had to be overcome. Thiswas done by developing the preionization process to create a relatively uniformdistribution of electrons prior to the main discharge and developing low-impedancecircuits to make very fast discharges possible. The first transverse discharge excimer lasersappeared in 1976 and 1977159 16,17,18. Electron beam pumping is still popular due to thehigh lasing efficiency achievable but discharge pumping is more convenient to use due to10Chapter 1: Introductionthe smaller size, simpler operation, and lower cost. A further possibility is the use of anelectron beam to stabilize a transverse discharge19; however, this will not be discussedfurther here.1.3 XeC1 Laser Chemistry and KineticsIn step with the development of laboratory excimer lasers was the development oftheoretical modeling of these systems. Excimer systems are very complex due to the largenumber of chemical species present. They are difficult to model because often over 50 rateequations are needed and many of the rate constants are not known accurately.Nonetheless, results that agree well with experiment can be obtained20. Most of theunderstanding of excimer chemistry is derived from theoretical modeling.As mentioned earlier, the XeC1 laser chemistry is fairly complex. The ground statecorresponds to a 1S rare gas atom covalently bonded to a 2P halogen atom. The weaklybound state is known as the X state and acts as the lower laser level. The upper laser levelis an ionicly bonded molecule consisting of a 2P state rare gas atom as the donor and a 1Shalogen as the acceptor. This is known as the B state; this molecule is known as an exciteddimer - hence the term "excimer" from the contraction.In the gas mixture are small amounts of the rare gas and the halogen donor. Themajority is a buffer gas such as helium or neon. In early XeC1 systems 02 was used as thechlorine donor; however, due to absorption at the laser wavelength by C12, efficiencieswere improved when HC1 was used instead. The active rare gas (krypton or xenon) canabsorb energy directly from the discharge; however, a buffer gas is necessary to preventthe glow discharge from becoming unstable. The light buffer gas molecules efficientlyabsorb energy from the accelerated electrons through elastic collisions and can transfertheir energy to the active species easily through Penning ionization or associativeionization.21 Helium is a good buffer gas; because of its low mass, it most efficiently1 1Chapter 1: Introductionabsorbs the electron energy. Other buffer gasses such as neon have a greater tendency toallow arcing due to its lower discharge impedance: less than half of that of helium.22Nonetheless, in many cases it might be preferable to use neon over helium because of thegreater lasing efficiencies possible20.The formation of XeC1* molecules can be summarized in figure (1.4). Depicted in thisdiagram are the dominant pathways for the formation of the excimer. It can be seen thatthere are at least two steps for the formation of XeC1*. Firstly, the excited or ionizedxenon atom has to be formed, collisionally with electrons or through Penning ionization bya metastable He* atom:Xe + e- —> Xe* + e-^ (1.4)OrXe(or Xe*) + e- -3 Xe+ + 2e^ (1.5)orXe + He* —> Xe+ + He- + e.^ (1.6)In this step, the electron distribution function is very important. Electrons gain theirenergy directly from the electric field between the electrodes. The electrons undergoelastic collisions until they reach the ionization potential or the first excitation energy ofone of the gas elements. They then loose their energy through an inelastic collision. The12Chapter 1: Introductionmetastable excited states play a large part in these excitation processes because of theirlarge cross sections and high concentrations.After the xenon atom is excited to a metastable state or ionized, the excimer can beformed through three pathways. The first involves a reaction with the excited HC1molecule,Xe* + HC1(v) --> XeC1* + H,^ (1.7)the second by ion- ion recombination with Cl-,Xe+ + Cl- + X --> XeC1* + X,^ (1.8)or the third in two steps,Xe+ + 2Xe --> Xe2+ + Xe^ (1.9)orXe+ + Xe + He --> Xe2+ + He (1.10)thenXe2+ + Cl- XeC1* + Xe.In reactions (1.8) to (1.10) it can be seen that a third, neutral atom, is required to removeexcess kinetic energy from a newly formed molecule.These reactions have been studied a great deal through computer modeling. In orderto model the excimer discharge, tables of collision cross sections are needed. These aretabulated in many sources as for example by Maeda et a/.21 For the calculation of electronexcitation or ionization rates, it is necessary to calculate the electron velocity distributionby finding a self consistent solution to the Boltzmann equation:— • — — —af^af^af (afat^ar^av^at)^. e rates for ion-ion recombi+ v^+ a^= Th nation can be determinedby Flannery's equation23 which utilises quasi-equilibrium statistics.It can be seen from equations (1.8) and (1.11) that the rate of formation of Cl- isessential to the efficient operation of the XeC1 system since it controls the rate ofcollision13Chapter 1: Introductionformation of the excited dimer. Cl- production occurs through the electron attachmentreaction,HC1 + e^H + ...(1.12)This reaction limits the pulse duration since the HC1 is not regenerated in the time scale ofthe pulse. Other factors prevent one from simply increasing the concentration of HC1 toprovide sufficient Cl- atoms25.The role of HC1 in the onset of arcing can also be seen. Arcing is thought to start witha HC1 local density less than the average for the system. Since the electron attachmentreaction (equation (1.12)) is the primary mechanism for electron loss, a paucity of HC1means that the density of electrons will increase over the average. This increase reducesthe local resistivity causing more current to flow locally thereby increasing ionization. Inthis way a positive feedback loop is established which triggers arcing24. This also indicatesthe importance of uniform preionization so that this feedback loop is not started.142 TheoryAll discussions in this thesis will utilize the rationalized MKS system of units.2.1 Definition of Electron TemperatureFor a system of particles in equilibrium, undergoing elastic collisions, a temperaturecan be defined by the expected Maxwellian velocity distribution, fo(v),(3V2 ) (^--^fo (v) = exp ---,7 ica 2 ) 2,^ ...(2.1)a'or1fo (v) = exp( mv2  )(2nkBTrikBT^, m )where v is the particle speed, it is the mean thermal speed, and T is the definedtemperature. Due to the large mass difference between electrons and ions, the electron-ionmomentum transfer cross section is very small. This means that the different species in theplasma can have different effective temperatures for short time periods.The effects of inelastic and superelastic (where an excited species is de-excited by thecollision) collisions will make the distributions non-Maxwellian. This effect has beenshown in many theoretical studies2° but has been shown to be less important inexperimental studies3,2° of excimer laser systems.2.2 Thomson ScatteringThe theory of Thomson scattering has been comprehensively outlined in Sheffield25,only a brief outline of the relevant points will be presented here. In this derivation, we arelaying to find the wavelength distribution of monochromatic light scattered from theelectrons in a plasma. To do this we will first find the radiation emitted by an electron15ScatteringVolumeFigure (2.1) Scattering DefinitionsChapter 2: Theoryaccelerated to a low velocity. Then we will assume that the charges are being acceleratedby the probe laser beam. Lastly we will find the effects of electron correlations and theelectron distribution function on the scattered spectrum.Taking Maxwell's equations and combining we get,1 a 2Ei^a JVx(VxE)+ 2^ ...(2.3)m 2 .= 110 at •cIf we take J to be the motion of single chargesJ = qv(t'),^ ...(2.4)where t' is the retarded time given byR' ... (2.5)and R' is the distance from the charge to the observer at time t'.Solving equation (3) we getE(R' ,t)^q^(E. -13(1-'))(1 -02 (tI)) ^q^xicys^)) x r.}1=^^4n eo^(1—g • 13(0) 3 R /2 +^eoc (1^• r3(0)3R'...(2.6)where 0 a: -- ^g is the unit vector from the charge to the observer (see figure (2.1)).For a low velocity charge, where 13«1, equation (2.6) reduces toE(R,t) = q^x(i x0(t))47c cocRAssuming that the incident radiation can be described asEi = Eo cos(k • r — co it)^ ...(2.8)and the position of the charge can be described asr(t') = r(0) + vt%^ ...(2.9)16Chapter 2: Theorysolving the equation of motion for an electron in the absence of a magnetic field,In* = qE„ and substituting into equation (2.7) gives the scattered electric field,Es (R,t) = q^^[‘i x(g xEio)]cos[k, -R—o)st — (k, —ki )- r(0)], ...(2.10)4n eocmRwhere cos = co (1^.13) is the Doppler shifted frequency and is the unit vector in the1(1—g. 0)direction of the probe beam propagation. It should be noted here that because of the massterm in the denominator, scattering from ions can generally be neglected. The scatteredwavenumber and frequency can also be found from the usual conservation of energy andmomentum arguments: ±cos = co, + coi and ±k, = ke + ki. In the case we are interestedin, and E0 are orthogonal therefore the X ( X E0)] term becomes -Eo.Given the radiated electric field from one electron, we can find the total radiatedpower simply by taking the vector sum of the contribution from each electron.dP, = eo c 2R 2 (/^/Eis )^ ...(2.11)c/CI^j=1^1=1which can be written asdP^22 [s = eoc R NE,2  + N(N —1)(E • Ei) jwi^ ...(2.12)where N is the number of electrons in the scattering volume. Obviously, the first term isthe power from individual electrons and the second term is the power from correlations ofelectrons.The Debye length, XD, is a very important parameter for plasma physics because it is thedistance over which charges will act collectively. At distances greater than the Debyelength, the relatively long range Coulomb interaction can be neglected due to shielding bysurrounding charges. If we are irradiating the plasma with a light beam that has awavelength that is much smaller than the Debye length then any correlations between thev( ET  )12electrons will not be apparent since the electrons are seen individually. XD = 2e newhere E is the dielectric constant and Ile is the electron density.17Chapter 2: TheoryA useful parameter is a E (kXD)-1. For incoherent scattering, a <<1 and scatteringis from individual electrons. In our case, we have n, = 3 x 1015cm-3,T, = 2 eV, 0 = 179.5°,X = 532 nm which gives an a of 0.22: definitely incoherent scattering. In furtherdiscussions, we will neglect the second term of equation (2.12) so the power is just thesum of the powers from the individual electrons.From an experimental point of view, we are interested in radiation scattered in acertain frequency range through a certain solid angle for the entire scattering volumedP,^eoc2R2N icivf (v)Es2(R,t)5((0s —cot —v •k).^...(2.13)The Es2 term can be brought out of the integral and the equation rearranged so that2Ps cko c/52 = q^NE.0 f dv f (v)5(co —wi —v • k)ckos^...(2.14)cocm)followed by2Ps do.), dfl = N( q^^Ei20 f(-Q2-)4t 0cmJ^k k...(2.15)This means that the reflected power at a given wavelength is simply proportional to thecorresponding velocity distribution. Knowing that the classical electron radius, ro is43r Eomc2 , and assuming a Maxwellian velocity distribution,2 tka 1 3/2 do)Ps dws d2= Pioro2 11,1, exp( :2a2 Y-2/^s di-2^...(2.16)Since k = ks ± ki, by the cosine rule, k2 = k k? —2k„ki cos 0 where 0 is thescattering angle. If we assume that I << 1 then k2 = —2k1 cos0 andC( co )2kac 2& 2 0^ .4a2X2^Putting this into terms that we can relate to experimentally,, sin 2(—2■ps (R, xs )dxs^PironeLcdS2 2^E exio p2772asin(-92)c2 Ax2^ dX4a2X,i sin 2 (9^Xs■ 2))...(2.17)dt2^218Chapter 2: TheoryTaking the logarithm of both sides, we can easily see that if In Ps is plotted against AX2C 2 C 2 Methen the slope would be^a or c^h^and the electron4a 2 X2i sin 2 (-iu)^8kBT,A,21 sin ( ‘±2)temperature can be readily obtained from the slope.19Chapter 2: Theory25J. Sheffield, Plasma Scattering of Electromagnetic Radiation (Academic, New York, 1975).203 Apparatus and SetupThe contents of this chapter will be confined to a physical description of theapparatus with very few details of the electronics. The following chapter will describethe precise synchronization of the components.3.1 OverviewAn overview of the apparatus is depicted in figure (3.1). The short seed pulse isproduced in the Spectra Physics oscillator and amplified by the ContinuumRegenerative Amplifier, injected through the excimer laser, and absorbed by a beamdump. The light scattered from the excimer discharge volume is collected with a largelens, spectrally dispersed in the spectrometer, measured, and analysed.3.2 Excimer LaserThe excimer laser in this experiment was constructed by Elezzabi for his M.Sc.thesisl. The description of this laser will be divided into sections discussing, in turn,the laser body, the circuitry, and the operation.3.2.1 Excitner Laser Body ConstructionThe glow discharge takes place between two parallel brass electrodes with amodified Chang profile26, 35 cm long with a 1.5 cm gap. For the experiments, thechamber was filled with a 5.3 atm mixture: [He]:[Xe]:[1-1C1] = 99.63 : 0.12 : 0.25. Thecharging voltage of the system was between 15 and 22 kV and the operating pressureswere between 3.1 and 5.6 atm. To reduce optical losses, the end windows were set atthe Brewster angle (OB = 57.10) for the UV light at 308 nm: the excimer lasing20Chapter 3: Apparatus and Setupwavelength. It was estimated that the glow discharge was approximately 0.5 cm wide,giving a discharge volume of approximately 26.25 cm3 .The gas was preionized by two rows of spark preionizers, placed half-waybetween and to the sides of the two electrodes (see figure (3.2)). The preionizers weremade by fitting 17, 2.5 cm long stainless steel tubes on a 5 mm diameter glass rod. Theends of the tubes were cut at 30° and spaced 1 mm apart so that when a large potential21Chapter 3: Apparatus and Setupdifference was applied to the ends, sparks would appear at each of the gaps. Many ofthe high energy photons produced in the sparks would be absorbed in an ionizing gasatom transition. This provided the necessary free electrons to enable a uniformdischarge. It was roughly estimated that such preionization produced an initial electrondensity of 107 cm-3.22Figure (3.3) Simplified LaserDischarge CircuitChapter 3: Apparatus and SetupThe laser was enclosed in a grounded fine copper mesh to reduce radio frequencynoise produced by the discharges.When this system was operated as a laser, a 100% reflecting mirror would beplaced at one end of the cavity and a quartz flat would be placed at the other. Themirror and the flat would then be aligned using a HeNe laser.3.2.2 Excimer Laser CircuitAs mentioned earlier, a fast discharge circuit isnecessary to avoid arcing in the discharge. For thisexcimer laser, a LC double inversion circuit is used.This circuit has the advantage of requiring only onespark gap and a power source supplying only onepolarity. This increases the reliability and reduces thetiming jitter of the circuit. This is also a relatively fastcircuit compared to other high speed dischargecircuits27.A diagram of half the circuit is shown in figure (3.3). Here, R1 is the spark gap,R2 is the main discharge volume, L1 is the circuit inductance and L2 is the laser headinductance. C* and C** are the energy storage capacitors. When the spark gap (R1) istriggered, current flows from the capacitors through the inductance and spark gapuntil the voltage across the main discharge (R2) is greater than the breakdownpotential of the gas. Then the main gap breaks down and current 13 starts to flow. Theinductance L1 from figure (3.3) is simply the loop inductance: calculated to be 27 nH.The discharge resistance (R2) varies with operating conditions but has been found tobe in the order of 0.5 O.23Chapter 3: Apparatus and SetupA complete electronic diagram, including the preionization circuit, is shown infigure (3.4). It can be seen that the preionizers are capacitively coupled to the maincircuit. This prevents sparking from the main electrodes to the preionizers by keepingthe voltage of the preionizers halfway between the electrode voltages. The preionizercapacitance was provided by two, 2.7 nF capacitors in parallel for each preionizationrod.To minimize the circuit inductance, the 24, 2.7 nF doorknob capacitors aredivided into four equal groups as shown in figure (3.4). The high voltage sides are24Chapter 3: Apparatus and Setupconnected to two large brass plates (21.1 cm x 33.2 cm x 0.65 cm).The spark gap electrodes are made of brass mounted in a Lucite chamber. Thedischarge was triggered by a -10 kV trigger pulse to a triggering pin located at thecentre of the cathode. This pulse was formed by a 4:1 step up transformer and aEG&G Krytron unit. Pressurized dry air flowed through the spark gaps to control thebreakdown voltage and to remove ozone. The air pressures and flow rates werecontrolled by manipulation of the air bottle regulators and small valves contained in theair exhaust lines.3.2.3 Excimer Laser OperationWith careful adjustments of the spark gap pressures and the voltage, the dischargejitter of the excimer laser could be kept under 5 ns. In other words, the standarddeviation of the time differences between the trigger pulse and the establishment of thedischarge in the excimer was less than 5 ns. Usually such a low jitter was notnecessary. It was easiest to take measurements over the entire pulse (-60 ns) not byprecise variation of the relative timing of the injection laser pulses, but by letting thenatural variation in timing randomly sample the pulse with each shot.The quality of the glow discharge could be monitored visually through a Plexiglassheet which filtered out ultraviolet light; any arcing would be readily apparent due toits brightness in relation to the glow discharge. Arcing was also easily detected by theirregular current signal that it produced.To stop persistent arcing a sequence of procedures was followed until the arcingdisappeared. Firstly, the spark gap pressures would be lowered until the pressure wasjust high enough to prevent a breakdown; secondly, the excimer body would beevacuated for a number of hours in an attempt to remove any impurities present andthen the cavity would be refilled with fresh gas; thirdly, if the previous steps didn't25Chapter 3: Apparatus and Setupwork, the chamber would be filled with helium and the electrodes would beconditioned by using a function generator to repetitively trigger discharges for anumber of hours; as a last resort, the electrodes would be manually cleaned and thesystem would be carefully examined for any air leaks.The gas recovery time for this laser was approximately 30 s. Any attempts atincreasing the repetition rate would, after a few shots, result in increased jitter andgreater chance of arcing. The gas mixture would have to be replaced every one or twodays due a buildup of the gas impurities.3.3 Injection Pulse ProductionThe probe pulse was produced by amplifying 5% of a —100 ps —100 nJ pulse froma Spectra Physics model 3800 Nd:YAG (Nd3+ doped Yttrium Aluminum Garnet) laserwith a Continuum RGA regenerative amplifier.3.3.1 Nd:YAG OscillatorThe laser oscillator utilized a Nd:YAG rod in the centre of a symmetrical 1.8 mlong cavity. The mode was controllable by a series of removable, 0.8 to 1.5 mm,apertures placed near the YAG rod. The quartz acousto-optic mode locker was placedinside the cavity near the output coupler. The acoustic driving frequency wasapproximately 41.1 MHz. The resultant pulse frequency was 82.2 MHz with a pulselength of approximately 100 ps. We didn't have the necessary apparatus to measure thepulse length more accurately. The average output power was approximately 12 W. Adielectric beamsplitter was used to split off 5% of this beam to be amplified in theregenerative amplifier.26WPM8SHGL2M7/^M5^A2M6PC2Al75% IM2I—BDI 532 nmWPR1 M3PC1 P2R2M4P1M1WPLockedPulsesCWMode-M1 to M7 - 100% MirrorsPl, P2 - Dielectric PolarizersLi - Diverging lens, -92 mmAl- 0.65 mm ApertureWP- Half Wave PlatesM8 - 532 nm Dielectric MirrorSHG - Second Harmonic GeneratorL2 - Converging lens 155 mmA2 - 3.5 mm Soft ApertureBD - Beam DumpR1 - Nd:YAG Amplifier Rod 6 mm diameter, 115 mm longR2 - Nd:YAG Amplifier Rod 9 mm diameter, 115 mm longFigure (3.5) Diagram of the Regenerative AmplifierChapter 3: Apparatus and Setup3.3.2 Regenerative AmplifierIn a regenerative amplifier, a seed laser pulse is switched into a cavity, makesmultiple passes through the gain medium, each time along the same path, and isswitched out. The layout of this regenerative amplifier is shown in figure (3.5).The pulse train was accepted into the regenerative amplifier through a Faradayisolator which prevented any feedback into the oscillator. At the mirror Ml, the pulsehas a vertical polarization. It can be seen that if the Pockets cell has no applied voltage,the two passes through the following half wave plate will cause the vertically polarizedpulse to be rejected by the polarizer Pl. However, if the Pockets cell rotates thepolarization such that after two passes it has rotated into the horizontal direction andthe wave plate rotates it into the orthogonal polarization, the pulse will be acceptedinto the cavity.27Chapter 3: Apparatus and SetupAt appropriate times the first Pockels cell was switched on and quickly off so thata single pulse would be accepted into the amplifier cavity. The pulse was thenamplified with each pass through the YAG rod until the gain medium was depleted.The pulse was then switched out by the second Pockels cell and sent through anadditional single pass Nd:YAG amplifier and a frequency doubling KDP (PotassiumDihydrogen Phosphate: KH2PO4) crystal. A dielectric mirror was used to separate the532 nm light from the residual 1064 nm light. The 1064 nm beam was directed into abeam dump since it was not needed. Green light pulses of up to 90 mJ could beproduced at 10 Hz. The power was measured using a Scientech Model 380101 EnergyDetector. This detector had an integration time of over 10 sec, which made themeasurement of single pulses impossible. 25% of this beam was removed from thetotal beam using a dichroic beamsplitter and was used for other purposes.It should be mentioned that the power actually used for Thomson scattering wasconsiderably less than 75% of 90 mJ per shot for two reasons. Firstly, as will bedescribed later, the laser was operated in "fixed" mode giving a less than optimumoutput power: estimated to be 70 mJ per shot. Secondly, a significant proportion ofthe probe beam was lost in reflections and by absorption. The energy reaching thedischarge was approximately 40 mJ per pulse. Unfortunately, we didn't haveequipment to conveniently measure the energy of a single pulse so visual estimates hadto be made using comparisons to the full power pulses on Kodak burn paper.The quarter wave voltages for the Pockels cells were applied by two Marx banks.Each consisted of a series of capacitors in parallel which were charged up to 300 V.When the "fire" signal was sent to the Marx bank, the terminals were electronicallyreversed so that the capacitors were in series and 2700 V was provided across the endterminals with a rise time of less than 1 ns.28Brass Chamber^ /1/2" Quartz flatsNom-- .FromRegenerativeAmplifier To Excimer^ 1 I^  f=433 mm f=127 mmTo Vacuum PumpFigure (3.6) Telescope Arrangement for Injected PulseChapter 3: Apparatus and Setup3.3.3 Steering OpticsTo inject the beam into the excimer laser and produce the required beam size, thelaser light was passed through a telescope made by two lenses: the first of focal length433 mm and the second of 127 mm. To avoid creating an air spark (an electricalbreakdown due to a high optical electric field), the focal point was enclosed by a smallvacuum chamber as shown in figure (3.6).The second telescope lens was adjusted so that the beam entering the plasma wascollimated with a diameter of approximately 4 mm. This diameter was necessary as itwas suspected that a smaller one would create electric fields large enough to create agas breakdown in the laser discharge volume which would give false measurements forthe experiment.All beam steering was done with prisms as it was found that any metal coatedmirrors would be damaged. Careful alignment was necessary to ensure good beamquality in the plasma.3.4 Collection OpticsThe back scattered light was picked up by a large, 307 mm anti-reflection coatedlens with an f number of f/2.5 (the f number is defined as the ratio of the focal lengthto the aperture diameter). The lens was positioned so that the scattering angle was29Chapter 3: Apparatus and Setup179.5°. A field stop aperture and a vertical polarizer was placed directly in front of thelens. The aperture was usually opened to 24 mm. The function of the polarizer was toattenuate diffusely scattered light. This would improve the signal to noise ratio sinceany Thomson scattered light would be vertically polarized.The collected signal was imaged onto the entrance slit of a f13.6 Jarrel-Ash82-410 0.25 meter Ebert Monochromater with the exit slit removed to create aspectrometer. The dispersed signal was then magnified as necessary and imaged ontothe entrance slit of the streak camera array or the face of the array of fibre bundleswhich passed the signal through the walls of the screened room to thephotomultipliers.Various precautions were taken to reduce stray light. To minimize light reflectedfrom the laser body and electrodes, the beam was passed through a metal mask with a4 mm circular hole just before it was injected into the plasma. To ensure that the lightscattered from the mask was not picked up by the collection lens, the beam path fromthe last prism to the mask was enclosed in a beam tube: a length of PVC tube paintedblack. After passing through the scattering volume the beam was directed into a stackof razor blades painted black28. This beam dump was designed to absorb the incidentlaser beam with as little reflection as possible and to provide an optically blackbackground for the collection optics. It was placed about a meter behind the exitwindow of the excimer and behind another light baffle in order to minimize any lightscattered from objects behind the laser.The field stop aperture was taped to the front of the collection lens. The lens,entrance slit of the spectrometer, and fibre optic array were all enclosed in blackcardboard boxes so that no light could reach the photomultipliers except through thefield stop aperture. The entrance slit of the spectrometer was also masked so that onlythe image of the discharge passed through the slit and all else was blocked. It wasfound that a significant amount of light was reflected off the front window, scattered30Chapter 3: Apparatus and Setupfrom the excimer body, and reflected back into the collection lens. So a small beamdump was placed beneath the excimer laser body to intercept this light.The size of the expected Thomson scattered signal can easily be calculated. Byintuition we can writePs^2- = ro neLdn^ ...(3.1)where L is the scattering volume length. Assuming that the collection lens has anaperture of 24 mm diameter and is 2.3 m from the centre of the excimer laser, equation(3.1) gives a ratio of scattered power to incident power of 8.6 x 10-15. Assuming thatthe incident energy is 40 mJ, this means that the collection lens `will receive 92 x 103photons - sufficient for this experiment.3.5 Optical System AlignmentCareful alignment of the probe beam and collection lens was done to ensure thatthe entire discharge volume was in the field of view of the collection lens and that the31Figure (3.8) Collection Lens FocusingArrangementChapter 3: Apparatus and Setupscattering angle was as desired. Most of the alignment was done with the use of twometal masks. Each consisted of a piece of aluminum with a pair of 2 mm pinholesdrilled with their centres separated by 8 mm. These were placed just outside of theexcimer windows (see figure (3.7)).A HeNe laser beam was set up so that its beam would pass through a 532 nmdielectric mirror (transparent at the 632 nm HeNe laser wavelength) and follow thepath of the injection pulse. This beam could therefore be used for alignment. Similarly,another HeNe laser beam was sent the other way through the discharge volume, thecollection optics, thespectrometer, and to the imagingdevice as shown in the diagram.It was also necessary to focusthe scattered light onto theentrance slit of the spectrometerwith the collection lens. This wasdone by placing a mirror in frontof the excimer laser and placing aHe Geisler tube, orientedvertically, in the path of the probebeam, the same optical pathlength away from the collectionlens as the centre of the excimerwould be (see figure (3.8)). Theimage of the Geisler tube wouldbe focused onto the closed slit of the spectrometer by changing the position of thelens.32same as the angle of maximumknown as the blaze wavelength.Chapter 3: Apparatus and Setup3.6 Scattered Light DispersionAs previously mentioned, a grating spectrometer was used to spectrally dispersethe signal. The spectrometer was constructed in what is known as the Ebertarrangement shown in figure (3.9). The large collimating mirror shown has a radius ofcurvature of 25 cm.The grating had an area of 64 mmby 64 mm and had 1180 grooves/mm,the grooves being oriented parallel to thedirection of polarization.The intensity of a monochromaticsignal reflected from a grating can bedescribed as/(y2 ) = D(01,(p2,X) where M is the total number of rulings andp = d(sin 01 — sin cp 2 ) is the path differencebetween parallel rays reflecting off adjacent rulingsand D(01, (p2 , X) is the diffraction pattern of anindividual facet. 0 1, y 2, and d are defmed infigure (3.10) The facets of the grating are cut at anangle to the plane of the grating so that at aspecified wavelength the angle of reflection is thediffracted intensity. This particular wavelength isThe grating used was blazed for 600 nm. For aspectrometer, the resolution can be defined as the smallest wavelength difference, AX,am, Grating^iFocussingACollectionOptics^ Optics/Colliminating MinorFigure (3.9) Ebert ArrangementSpectrometer33Chapter 3: Apparatus and Setupsuch that when two monochromatic lines, differing by AX are input into thespectrometer, the widths of the diffracted lines are equal to the separation of theircentral maxima.It can be seen that as the number of grooves increases, the resolution will alsoincrease. For high resolution, the light incident on the grating must be highlycollimated. Therefore, to optimize the resolution of the spectrometer, two things canbe done:1. The f number of the signal can be matched to the f number of the spectrometer sothat the entire grating is covered;2. The entrance slit of the spectrometer can be narrowed down until the width ofthe transmitted signal is diffraction limited.Unfortunately, for most of these experiments, the full grating was not illuminated.In order to minimize stray light entering the spectrometer, the field stop aperture wasemployed to keep the f number of the collection lens large (f/12.8). This meant that theresolution obtainable was 3.5 times less than that theoretically possible: the gratingcould theoretically resolve lines 0.2 nm apart instead of lines 0.06 nm apart.The slit width used for most of these experiments was determined to be 320 Jim.Given a collimating mirror focal length of 25 cm, the width of this entrance slit wouldcreate an uncertainty in the incident angle, 0 1, of 1.3 x 10-3 rad which created anuncertainty in the diffracted angle, y 2, of 9 x 10-4 rad. The loss of resolution from thefinite width of the entrance slit was calculated to be much smaller than the loss ofresolution due to the finite number of rulings.3.7 Streak CameraA streak camera is a device which allows the high temporal resolution of a onedimensional spatial image. Light passing through a horizontal slit is imaged onto a34Chapter 3: Apparatus and Setupphotocathode converting an optical image into an electron image. This image isaccelerated and passed between parallel deflection plates as shown in figure (3.11). Asweep voltage is applied to these plates at an appropriate moment so that electronsarriving at different times will have different vertical displacements on the phosphorscreen and the horizontal image will remain unaffected. The phosphor image isrecorded by a video camera.The streak camera used was a Hammamatsu 979 Temporal Disperser attached toa C1000 video camera and a C1440 Frame Memory Image Analysis System. Five gainsettings were available. The speed most commonly used scanned the full screen in5.42 ns.The lens combination used between the entrance slit of the streak camera and thestreak tube was a 105 mm, 35 mm pair giving three times magnification. The streakcamera entrance slit was either put directly against the output aperture of thespectrometer to collect a higher intensity signal with lower resolution or separatedwith a magnifying arrangement to provide better spectral resolution but lowerintensity. Unfortunately, it was found that there was not enough dispersion to getuseful results when the camera was placed directly against the spectrometer.35Chapter 3: Apparatus and SetupWhile a spherical lens magnification arrangement would have been preferable to acylindrical one due to the simplicity of alignment one was found to be inadequate dueto the light lost at the streak camera entrance slit due to the magnification in thevertical direction.Using cylindrical lenses arrangement, as shown in figure (3.12), it was possible tohave a reasonable magnification in the horizontal plane while focusing all the light inthe vertical direction onto the horizontal entrance slit of the streak camera. The lensesused each had a focal length of 7 cm, a width of 5.5 cm, and a length of 22 cm. Thestreak camera was placed approximately 0.75 m from the spectrometer with the lensesplaced at the appropriate positions to achieve the proper focus.Two dimensional angular adjustments had to be made to all components sincethere was an internal misalignment in the spectrometer. This meant that the outputsignal was not traveling parallel to the table and the direction of dispersion was nothorizontal. These problems introduced distortion into the signal. It was discovered thatthe streak images obtained did not have a precisely vertical time axis and that as thestreak camera grating was rotated, the intensity of the signal would vary. It wasassumed that the variation wasn't significant on the wavelength scale of the dispersedimage.36Chapter 3: Apparatus and SetupIn order to reject light at 532.0 nm while accepting the spectrally dispersedThomson scattered light, for some preliminary experiments, a very thin piece of blacktape was placed vertically across the horizontal slit to block out the signal exactly at532.0 nm. In this way, the full intensity of the dispersed image could be imaged on thestreak tube with maximum gain without damaging the system. However, it was foundthat when all sources of stray light were successfully eliminated, the total signal wasnot bright enough to damage the camera so the mask was removed.Alignment, focusing, and calibration of the streak camera was done with a highlyattenuated HeNe laser passing through the excimer laser, collection optics, andspectrometer as described in section (3.5). Further focusing was done by focusing thespectrally dispersed image of a He Geisler tube onto the streak tube and watching theimage on the video display. The Geisler tube was positioned as shown in figure (3.8).With this experimental arrangement, the best time resolution attainable isapproximately 2.5 ns due to differing delay times introduced by the scattering volumelength. In practice, the time resolution would be a bit lower due to the width of thestreak camera entrance slit needed to get an appropriate intensity on the streak tube.The streak images were transferred to a personal computer and analysed. The analysisprogram was written by Y. Zhu with modifications by myself.3.8 PhotomultipliersThe three photomultipliers used were RCA 8575's operated with a -1700 V bias.It was found that the photomultipliers had to be operated inside the screened room toavoid the pickup of noise created by the spark gaps. To get good coupling betweenthe fibres and the photomultiplier tubes, the bundles were epoxied into a hole drilled ina brass plug. The plug and the fibres were then polished. The plug was then fitted intoa jacket so that the fibre abutted a Plexiglas window. The photomultipliers were37Chapter 3: Apparatus and Setupmounted in such a way as to abut the window from the other side as shown in figure(3.13). The image from the spectrometer was magnified nine times and imaged ontothe fibre optic array. The physical size of each fibre optic channel was 1.6 mm wideand 2.5 mm high. Due to the height of these channels and the use of the spherical lensfor magnification, the alignment problems experienced with the cylindrical lenses andthe streak camerawere not experiencedwith this arrangement.To reduceelectronic noise, thephotomultiplier tubeswere wrapped incopper sheaths whichwere kept at thecathode potential andall the tubes were enclosed in a larger copper sheath kept at ground potential.The photomultipliers were calibrated by measuring the voltage signals from theprobe light reflected off a white card placed at the same optical path length as theexcimer see figure (3.14). It was assumed that the light reflected off the card wasessentially monochromatic. Voltage signals were taken as a function of the wavelengthreading on the spectrometer. Care was taken to always turn the spectrometer dial inonly one direction to avoid backlash. The results are shown in figure (3.15). Using thisinformation, the wavelength separation between channels was found to be 0.6 nm. Abetter calibration was not performed because ultimately the calibration was notneeded.38Chapter 3: Apparatus and SetupUnfortunately, with thephotomultipliers, the best timeresolution attainable was about5 ns. This was not good enoughto distinguish the noise from theend windows from the signal as itwas possible with the streakcamera. The streak cameraimages showed that the intensityof the reflected signal from theend windows would besufficiently large to hide anyThomson scattered signal as willbe shown in chapter 5. Thetemporal resolution was estimated39Figure (3.16): Photomultiplier Voltage for a Single Photon.Chapter 3: Apparatus and Setupto be the full width at half maximum of the signal from a single electron emitted by thephoto-cathode which was on the order of 5 ns (see figure (3.16)).3.9 Electronic MeasurementsSpecial precautions had to be taken to counter the high level of radio frequencynoise created by the sparkgaps and the excimer discharge which affected electronicmeasurements and created false timing signals. The two Tektronics oscilloscopes werehoused in a screened room to avoid electrical interference. This room is constructedwith double walls of fine copper sheet connected to the ground potential at only onepoint. The power lines were passed through a noise filter and the coaxial signal cableshields were grounded at the wall. An electronic schematic diagram of themeasurement apparatus is shown in figure (3.16).40Chapter 3: Apparatus and SetupThe photomultipliers were particularly sensitive to noise. AS a result they had tobe housed within the screened room. As previously described, the output from thespectrometer was magnified onto an array of fibre optic bundles. The fibres were thenpassed through a small hole in the walls of the screened room and coupled to thephotomultipliers.To perform this experiment, it was necessary to measure the photomultipliervoltage signals, the current through the plasma, and a photodiode signal of theinjection pulse. Unfortunately, when the excimer laser was fired, the photodiode signalwas overwhelmed by the radio frequency noise created by the discharge.Arrangements could have been made to remove most of this noise but, due to the lackof positive results, it was not felt that this would be productive.The excimer discharge current was measured with a linear Rogowski coil. This.....•coil simply consisted of a copper wire wrapped around an insulator N times. When the41Figure (3.18) Rogowski Coil Current as Seen on OscilloscopeChapter 3: Apparatus and Setupcurrent passing through the plasma produced a changing magnetic field, the changingflux through the coil induced a proportional voltage in the coil as predicted byAmpere's law.The Rogowski coil was placed alongside the excimer body and under a row ofcapacitors so that the current flowing through the upper electrode could be directlymonitored. A sample of the Rogowski coil voltage is shown in figure (3.18).The photodiode was placed so as to pick up light reflected from one of the lensesof the telescope.Since we had only two oscilloscopes, we had to delay the signals from the two ofthe three photomultipliers and add them together at the inputs. One signal was delayedby 100 ns and another by 700 ns.424 Electronic Timing and SynchronizationThe timing of all the components is crucial to the success of this experiment due to the shortinjection pulse duration 100 ps) and the low repetition rate (---= 0.02 Hz). The injection pulsemust coincide with the excimer laser discharge and the measurement apparatus must be triggeredat appropriate times in order to capture the signal.This chapter will be organized into sections describing first how the components of theregenerative amplifier are timed to each other then how the excimer laser and the measurementapparatus are timed in relation to the regenerative amplifier.A non-commercial electronic delay box was used as the primary delay device. The delaycould be varied from 0.0 to 999.9 ps in steps of 0.1 ps. The delay box was triggered from thedifferentiated pulse from the flashlamp trigger pulse; the Pockels cells and excimer laser triggerpulses were produced at appropriate times later.4.1 Regenerative AmplifierIn the previous chapter, a simplified explanation of the operation of the regenerativeamplifier was given. In this chapter the electronic subsystems will be described. These can beconsidered as comprising of three subsystems, the Control Unit, the Power and Capacitor BankUnits, and the Laser Bench Unit. Each subsystem will be described in turn.The effect of the controls to be described was to keep the jitter in the timing of the greenpulse as seen on the oscilloscope below approximately 3 ns. However, this could have beencaused by jitter in the delay box or in the oscilloscope triggering circuit and was not necessarilyentirely due to the RGA.43Chapter 4: Electronic Timing and Synchronization4.1.1 Control UnitAt the heart of the regenerative amplifier is the Control Unit. As the name implies, theControl Unit controls the operation of the Power Units and Capacitor Banks. It sends chargesignals to the flashlamp Power Unit and sends a fire signal to the flashlamps as soon as thecapacitors are charged. These signals are sent at a predetermined rate between 5 and 10 Hz,triggered manually through two pushbuttons on the control panel, or triggered externally thoughthe "external connector". For this experiment, either the flashlamps were fired repetitively andautomatically ("fixed" mode) or controlled externally ("single shot" mode). Diagrams of eachmode are shown in figures (4.1) and (4.2). These diagrams will be gradually explained throughoutthis chapter.The first mode, dubbed the "single shot" mode, utilizes the external triggering circuitry of theregenerative amplifier Control Unit. The interface is a 9 pin external connector. Table (4.1) listsall the connections made. As can be seen, pin 2 is the only one used dynamically; a pushbuttonswitch is used to connect the high voltage from the power supply to the pin when a shot is44Chapter 4: Electronic Timing and Synchronizationdesired. The other pins are fixed either at 0 or 5 V. Pin 3 could be used to fire the flashlamps butis set so that they are fired as soon as an end of charge signal is received.The "fixed" mode was used to produce a maximum single shot power. In this mode, aninternal clock sent charge signals to the Power Unit at a fixed rate between 5 and 10 Hz. As soonas a end of charge signal was received from the Power Unit, a trigger signal was sent from theControl Unit to the Power Unit to fire the flashlamps (see figure (4.2)).45Chapter 4: Electronic Timing and SynchronizationPin 1: Power supply groundPin 2: Charge command 0 to 5 VPin 3: 5V. Fire commandPin 4: 5V. needed to provide an end of charge signalPins 5 through 9: not used.Table (4.1) External Connector Pin Assignments for"Single Shot" Mode4.1.2 Power and Capacitor BanksThe regenerative amplifier needs two sets of Power Units and Capacitor Banks: one for eachlaser rod. As mentioned before, the Control Unit sends identical charge and fire signals to thePower Units. As soon as the charge signals are received, the Capacitor Banks are charged to thepre-selected voltage of close to -1700 V. This voltage can be adjusted by a dial on the front panelof the Power Unit. Slight adjustments are occasionally needed to maximize the output power.Once the capacitors are charged, the end of charge signals are sent to the Control Unit.When the fire signals are received, fast -17 kV ionizing pulses are sent to the flashlampsfollowed by the flashlamp voltage pulse which has a duration of 180 !is (FWHM). At the sametime, a 24 V high to low signal is sent from the resonator Power Unit to the external electronicsvia a BNC output, denoted J5, which fires the Pockels cells after an appropriate delay. The 24 Vsignals were differentiated by a 0.047pF capacitor. This differentiated signal could be used totrigger the delay box.4.1.3 Laser Bench ControlsThe frequency scaler unit controls the timing of the Pockels cells in relation to an externalsignal and the seed mode locked pulse train. This unit takes the 82 MHz signal from the oscillator46Chapter 4: Electronic Timing and Synchronizationacousto-optic modulator and converts it to a fast TTL signal for its own use. When theappropriate trigger signal is received from the external electronics a switch is set so that the firstPockels cell is fired at the beginning of the next cycle and a 15V trigger signal is sent to a BNCport marked "sync out". At the same time a delay line is activated so that after an appropriatenumber of clock cycles, the second Pockels cell is supplied with the quarter wave voltage toswitch the amplified pulse out of the oscillator.The input radio frequency synchronization signal from the YAG acousto-optic modulator hadto have a peak to peak voltage of 3 V and the trigger signal from the delay box had to have anominal voltage of 15 V. A voltage divider was used to attenuate the 40 V delay box signal.Occasionally adjustments had to be made to the relative timing of the first Pockels cell to thesecond one in order to get the maximum possible power out.4.2 Delay BoxThe 24 V signal from the J5 output of the Power Unit was differentiated by a 0.047 m.Fcapacitor. This differentiated signal could be used to trigger the delay box. In the "single shot"mode the signal goes directly to the delay box as shown in figure (4.1). In the "fixed" mode, apushbutton switch is used to connect the trigger signal to the delay box when desired: as shown infigure (4.2).The 16 channel delay box was used to provide 40 V positive edge trigger signals to the restof the system. Signals from the delay box triggered the Pockels cells at the appropriate times aswell as fired the excimer preionizer and excimer main discharges. It also provided the appropriatetrigger signals for the oscilloscope. The delays used are indicated on figures (4.1) and (4.2).4.3 Streak CameraThe streak camera was triggered by the 15 V low to high "sync out" signal from theregenerative amplifier 50 to 100 ns before the first Pockels cell was fired. It was triggered with47Chapter 4: Electronic Timing and Synchronizationthis synchronization signal instead of with a photo-diode because the delay between the inputtrigger pulse and the streak camera operation was at least 28 ns. This would imply anunreasonable optical path delay. Adjustments to the timing of the streak camera in relation to thescattered signal was done by manipulation of a 0 to 32 ns delay box with increments of 0.25 nssupplemented by appropriate delay cables. To time the injection pulse and the streak cameraproperly, small targets were set up at the entrance and exit windows of the excimer; the reflectedsignals from which could be seen very clearly on the streak camera. Thus appropriate delayadjustments could easily be made.4.4 Measurement ApparatusIn order to see the measured signals at the appropriate times, a fourth channel of the delaybox was used to trigger both oscilloscopes at the same time. The BNC co-axial cables from thephotodiode and the Rogowski coil were made the same length so that accurate time comparisonscould be made. The diode was operated with a 27 V bias.485 Measurements and ResultsIn this chapter, the measurements taken by the streak camera and the photomultipliertubes and how they were analysed is discussed. This is followed by a section discussing allthe results as a whole.5.1 Streak CameraAt first, measurements were made with the streak camera. A representative exampleof a streak camera image is shown in figure (5.2). The density of shading is proportionalto the intensity of the signal. Indicated are the time and wavelength axes. The signal can beseen in the middle between the scattered signals from the two end windows. The datawere recorded as a 512 by 512 pixel matrix with each pixel having an intensity valuebetween 0 and 255.49Chapter 5: Measurements and ResultsThe position of the Rogowski coil current in relation to the probe pulse on theoscilloscope screen was recorded for each streak record so that the recorded scatteredsignal could be related in time to the excimer pulse. Similar data were also recorded forcases when the probe beam was fired without an excimer discharge. These data could beused as a background level.50Chapter 5: Measurements and ResultsTo analyse the raw data, the signal and background shots were compared. Figure(5.1) shows the data in figure (5.3) temporally averaged over 0.75 ns around themaximum intensity. Also shown is a similarly treated background signal.Since the intensity of the probe beam would have been slightly different for each shot,in analysis, the puls powers were all normalized to equal values. It is valid to simply scalethe signals since the scattered intensity for any effects that we have considered isproportional to the incident intensity. From figure (5.1), it can be seen that there is nosignificant difference between the two shots. Therefore any Thomson scattered signal ismuch smaller than the noise signal.Assuming an electron temperature of approximately 2 eV', the full width halfmaximum of the signal should have been approximately 3 nm. In figure (5.1), the spectrumhas a Full Width at Half Maximum (FWHM) of approximately 0.8 nm. Any signal that wewould be able to detect therefore would be the portion shifted by more than approximately0.5 nm. Since we could not find any signal at such shifted wavelengths and since thisstreak record was taken with maximum gain and maximum practical streak width, it wasdecided that the much more sensitive photomultiplier tubes would have to be used. Thephotomultiplier tubes had the additional advantage of being easily able to reject the noisesignal at 532.0 nm simply by tuning the spectrometer.The streak camera results were very useful in determining where the excess stray lightwas coming from as deduced from the timing information produced. From these results itwas determined that the majority of stray light was coming from the front and backwindows. In fact, over half the intensity reaching the streak camera was scattered lightfrom the windows. This could easily be seen in figure (5.3) which was made by taking thesignal from figure (5.1) and integrating it over the wavelength axis to find the intensityreaching the camera at any particular time.51Chapter 5: Measurements and ResultsSince the Rayleigh scattering cross section in relation to the Thomson scattering crosssection is well known for helium, an estimate of the noise expected can easily be made.DeSilva and Goldenbaum29 give aT/GR=2750 which leads to an estimate of the ratio of thescattered intensity to the incident intensity of 1.4 x 10-7 which is eight orders of magnitudeabove the Thomson scattered signal.Rayleigh scattering was probably the cause of much of the noise seen. Rayleighscattered light originates from oscillating atomic dipoles excited by the incident radiation.Any wavelength shift arises from the motion of atoms in much the same way as thewavelength shift occurs in Thomson scattering; however, at room temperature, theRayleigh scattered signal is essentially monochromatic for monochromatic probe light.1—^1;0.9 — I Pli08—^I; It0.7 — 1.iI •10.6 — r... I.^I64 0.5 — IF^!0)^1.141^•I4 0.4 --al: % I •0.3 a z^I^Om0.2^il FL :^•^•I^160 . 10 ^VA i t 9 Pi%— as^I___Awmannirl.^t rr^difill11-10^1^2^3^4^5time (ns)Figure (5.3) Analysed Streak Camera Image52•I.20 4 6• •• rpa11.1.•°AI •lp..1111111111111111111111111ITI rprrrU111111111111111111111111••• MI••lieurdrigmarmarI.Time (ns)Figure (5.4) Background and Signal Time Signals to show Rayleigh Scattering• Background•^SignalChapter 5: Measurements and ResultsThis was definitely shown in the experimental results. There was a great difference innoise signals between when the excimer body was filled with gas and when it wasevacuated. Figure (5.4) depicts the signal shot shown previously compared to abackground shot taken when there was no gas in the excimer laser body. To obtain thisfigure, first the two images were time shifted so the front window reflection peakscoincided then these peaks were normalized to the same value.5.2 Photomultiplier TubesA typical voltage signal from the photomultiplier tubes, as seen on the oscilloscope, isshown in figure (5.5) for when the excimer is firing, and in figure (5.6) for when it is not.53Chapter 5: Measurements and ResultsIn general, in order to see if any signal could be distinguished from the noise, ratiosbetween these voltages were taken and compared statistically. It was necessary to useratios instead of the absolute voltages as, when the excimer was firing, the probe pulsepower could not be measured with the present arrangement. There were no significantdifferences found for any of the data sets: each consisting of 20 to 40 shots. Attemptswere made using the laser gas mixture as well as pure helium at pressures of 3.7 atm andFigure (5.5) Typical Photomultiplier Tube Signals with an ExcimerDischarge5.3 atm. There were no significant differences found between these cases either. However,there were large differences when there was no gas in the excimer compared to when therewas gas, again confirming the suspicion that the majority of the noise was from Rayleighscattering.54Chapter 5: Measurements and ResultsAlong with the photomultiplier data with each shot, measurements were made of thedifferentiated excimer current when the laser was firing and the diode voltage when it wasnot. It should be noted that the excimer current traces within each data set were identicalwithin the experimental uncertainty (recall that the diode signal was not usable when theexcimer was firing).5.3 DiscussionThere were a few problems with the experiment that made measurement of theelectron velocity distribution impossible: poor resolution, large noise signals at the laserfrequency, and possibly laser induced heating.5.3.1 ResolutionAll of our results were affected by a lack of resolution. In order to determineFigure (5.6) Typical Photomultiplier Tube Signals Without anExcimer Discharge55Chapter 5: Measurements and Resultssomething about the resolution of the system, we can look at the measured widths of whatshould be essentially monochromatic signals: the Rayleigh scattered signal from theexcimer gas and the scattered signals from the front and back excimer windows. Suchsignals are shown in figure (5.6) along with a signal that should show a Thomson scatteredcomponent. From the spectral width of the signals reflected off the front and backwindows the resolution problems are readily apparent. The signals which are supposed tobe essentially monochromatic - all except the one marked "signal" - have approximatelythe same widths as the signal shot. Furthermore, while the signal should be symmetric,there is a large "tail" on the negative wavelength shifted side of the spectrum.Following is a list of a few possible explanations for the lack of resolution: only asmall area of the grating was illuminated, the spectrometer entrance slit was opened toowidely, scattered light from grating and magnification system was picked up by themeasurement apparatus, or the dispersion system was slightly out of focus. Each probablycontributes significantly to the total problem.As mentioned in section (3.6), to mask out stray light, the field stop aperture in frontof the collection lens was closed down to 24 mm. This meant that the f number (ratio offocal length to aperture diameter) of the collection lens was much greater than that of thespectrometer and therefore that the area of the grating used was much smaller than thetotal area. The reduction in the number of rulings used led to a theoretical resolutionreduced by 3.5 times from the maximum possible resolution. This meant that the minimumtheoretical resolvable wavelength difference would be approximately 0.1 nm. From figure(5.6) it is obvious that the resolution is much worse that this.The second problem resulted from the fact that in an effort to collect the maximumscattered signal possible, the entrance slit to the spectrometer may have been opened toowidely. However, from analysis, it can be shown that this wasn't the case. From thecalibration data in figure (3.12), it can be seen that the FWHIM of the photomultiplierchannels was only 0.5 to 0.7 nm and from figure (5.1) it is 0.8 nm. Based on an entrance560.2-3 -2 2 3•0.80.4••• • • • • • • • • • • • • •°— Signal^ Rayleigh Scattered•— Front Window^ Back WindowWavelength Shift (nm)Figure (5.6) Signals Showing Resolution ProblemsChapter 5: Measurements and Resultsslit width of 320 p.m, the theoretical FWHM of a monochromatic signal would be less than0.01 nm so the large entrance slit width does not cause a resolution problem.Thirdly, there could have been a significant amount of scattered light from the gratingand optics of the dispersion system. Again, this was not seen to be a major problembecause the calibration measurements seemed reasonably good although not near whatwas theoretically possible.Lastly, the focusing elements of the system could have been slightly mispositionedleading to less than optimal spectral resolution. However, this should have not been aproblem at all because all of the optics were re-focused on a regular basis.57Chapter 5: Measurements and Results5.3.2 NoiseThe second major problem was excessive stray scattered light at the laser frequency.It was this large amount of noise in relation to the Thomson scattered signal that made thelack of resolution such a problem. If the resolution was better, the stray light could simplyhave been excluded by masking it out.The noise was scattered from a few sources: from the windows, the excimerelectrodes, the excimer gas through Rayleigh scattering, or from dust.29 A lot of effort wasput into alleviating these problems as has been previously discussed in the apparatuschapter; however, large Rayleigh scattering signals will always be present and thereforeimprovements will have to be made to the resolution of the system. Further ways ofreducing the noise signal are discussed in the following chapter.5.3.3 Laser Heating EffectsLaser induced heating or a laser induced plasma could have affected all the results.Regarding figure (5.3), while the signal should come evenly from the entire scatteringvolume, the signal is entirely produced from the front of the excimer body. A possibleexplanation for this is that there was a laser induced plasma at the front of the excimerbody. The free electrons would have presented a large scattering cross-section for theinjected beam. However, the potential for such a problem was recognized shortly afterphotomultiplier tube measurements were started and hopefully corrected for by expandingthe beam. There was no visible evidence of such heating; it would be expected that thebreakdown would have been visible when looking through the excimer gas through a532 nm filter but it was not.It could also be possible that there was another laser heating effect similar to the onenoted by Uchino et a13. They reported that it was not possible for them to performmeasurements on plasmas containing xenon due to unexplained laser heating effects. It58Chapter 5: Measurements and Resultswas not known if our experiment suffered from a similar problem. Again, possiblesolutions to this problem are discussed in the following chapter.5.3.4 ConclusionsIn conclusion, the major problem affecting this experiment was the lack of spectralresolution. If this problem had been solved, it is felt that it would have been possible tomeasure the electron velocity distribution.In addition, the major problem affecting the photomultiplier tubes was that the timeresolution was not good enough. The considerable noise from the end windows couldeasily mask any signal given the lack of resolution.The main problem affecting the streak camera measurements was that the gain wasnot high enough to measure the Thomson scattered signal at the extremes of the shiftedwavelength spectrum.596 Future WorkIf this experiment were to be attempted again, it would be worthwhile to change theexperimental arrangement to mitigate some of the problems mentioned in the previoussection.The biggest improvement might come about if a different scattering geometry wasused. From a general survey of the literature, it seems that most experiments of this typehave been performed with a scattering angle closer to 90°. This would greatly reducedirectly scattered stray light. Such a change would also increase the time resolution of theexperiment due to the shorter scattering length and resulting smaller range in round triptimes. In order to do this, modifications would have to be made to the excimer laser bodysuch as inserting a window in the side.In an attempt to reduce stray scattered light, it would be beneficial to see if scatteredlight from dust in the chamber contributed significantly to the total noise level. This couldbe done by investigating how the signal varies with pressure. Any signal from dustparticles wouldn't vary proportionally with pressure but would decrease with increasedsettling time between shots. To minimize any signal from dust in the chamber, it would benecessary to wait a significant amount of time between shots for all dust to settle down.In order to get a measurable electron temperature, the helium buffer gas should bereplaced with neon20,30. This would result in a higher electron temperature so that thespectrum of the Thomson scattered signal would be shifted farther out of the range of thenoise spectrum.Prior to another similar experiment, it would be worthwhile to investigate why amixture containing xenon would suffer from the laser heating effect reported by Uchino etal. It would be necessary to look at the absorption spectrum of xenon atomic states andmolecules. To see if this experiment experienced such an effect, it would be necessary to60Chapter 6: Future Workperform the same type of analysis they did: it should be seen if the scattered signal poweris proportional to the probe laser power as is desired.To improve the experiment so that results would be more accurate and easier toobtain, a number of steps should be undertaken.Firstly, it would be necessary to measure the intensity of the probe beam with everyshot. That means it would be necessary to either screen the noise from the excimer laserbetter or to put the diode inside the screened room.To facilitate the rapid collection of data more photomultiplier tubes should be used.This would allow a sufficient number of data points to find an accurate distribution withevery shot. Presently, the grating would have to be rotated periodically so that a numberof data points sufficient to calculate the distribution could be found. This method hasdefects arising from the need to take many shots and average the data. The current fibreoptic array could possibly accommodate up to seven photomultipliers; however, anincreased spectral dispersion would be necessary.Even more desirable would be to decrease losses in the optical path before theexcimer laser. The additional power in the signal would allow the streak camera to be usedinstead of the photomultipliers. The streak camera has advantages over thephotomultipliers in that noise from the front and back windows does not interfere withmeasurements because it is temporally separated from the signal and that measurementsdone with the streak camera are easy to analyse because of the continuous wavelengthspectrum. To increase the probe power, all the prisms could be replaced with dielectricmirrors and the lenses and flats could be anti-reflection coated.To address the resolution problems mentioned in the previous chapter, the currentspectrometer used should be replaced with a larger one with more resolution.Alternatively, a second spectrometer should be placed behind the first to additionally rejectany stray light. Hopefully this would also alleviate the problem shown in figure (5.6)61Chapter 6: Future Workwhere the spectrum isn't symmetric around the centre. It was never determined why thiswas so but it must have had something to do with the alignment of the spectrometer.It might also be of interest to calibrate the photomultipliers with an absolute intensityso that the electron temperature could be measured at the same time as the electrondensity. A common method for doing this is to calibrate against the Rayleigh scatteredsignal. As previously mentioned, the electron density is an important parameter formodeling excimer lasers.With such improvements, a successful experiment could certainly be performed.62References1. A.Y. Elezzabi, M.Sc. thesis, University of British Columbia (1989).2. H. Yamakoshi, M. Kato, K. Uchino, T. Iwata, M. Masuda, K. Muraoka, M. Maeda, M. Akazaki; Jap.Journ. Appl. Phys. 28, L1589 (1989).3. K. Uchino, Y. Kubo, K. Muraoka, T. Sakoda, H. Yamakoshi, M. Kato, A. Takahasi, M. Maeda; J.Appl. Phys. 70, 41 (1991).4. P. Millonni and J. Eberly, Lasers (John Wiley & Sons, Toronto, 1988).5. A. Yarriv, Quantum Electronics (John Wiley & Sons, Toronto, 1989).6. H. Griem, Plasma Spectroscopy (McGraw-Hill, Toronto, 1964).7. G. Fiocco and E. Thompson, Phys. Rev. Lett., Vol 10, 89 (1963).8. E. Thompson and G. Fiocco, MIT Research Laboratory of Electronics Quarterly Progress Report No.69,74 (1963).9. M. Krauss and F. Mies, Excimer Lasers, ed C. Rhodes (Springer-Verlag, Berlin, 1983).10. J. Ewing and C. Brau, Phys. Rev., Vol Al2, 129 (1975).11. J. Velazco and D. Setser, Jour. Chem. Phys., Vol. 62, 1990, (1975).12. S. Searles and G. Hart, App!. Phys. Lett., Vol. 27, 243 (1975).13. J. Ewing and C. Brau, Appl. Phys. Lett., Vol. 27, 350 (1975).14. E. Ault, R. Bradford, and M. Bhatunik, Appl. Phys. Lett., Vol. 27,413 (1975).15. R. Burnham, N. Harris, and N. Djeu, Appl. Phys. Lett, Vol 28,707 (1976).16. C. Wang, Appl. Phys. Lett, Vol 29, 103 (1976).17. R. Burnham, E. Powell, and N. Djeu, Appl. Phys. Lett, Vol 29, 30 (1976).18. W. Sarjeant, A. Alcock, and K. Leopold, Appl. Phys. Lett., Vol. 30,635 (1977).19. J. Mangano and J. Jacob, Appl. Phys. Lett., Vol. 27, 495 (1975).20. M. Maeda, A. Takahashi, T. Mizunami, and Y. Miyazoe, Jap. J. Appl. Phys. Vol. 21, 1161 (1982).21. C.A. Brau, Excimer Lasers, ed. C.K. Rhodes (Berlin: Springer-Verlag, 1983).22. S. Watanabe and A. Endoh, App!. Phys. Lett., Vol 41, 1(1982).23. F. Flannery and T. Yang, Appl. Phys. Lett., Vol. 32, 327 (1978).24. J. Coutts and C. Webb, Jour. Appl. Phys., Vol 59, 704 (1986).63References25. J. Sheffield, Plasma Scattering of Electromagnetic Radiation (Academic, New York, 1975).26. T.Y. Chang, Rev. Sci. Inst., Vol. 44,405 (1973).27. H. Houtman, A. Cheuck, A. Elezzabi, J. Ford, M. Laberge, W. Liese, J. Meyer, G. Stuart, and Y. Zhu,Rev. Sci. Inst., Vol. 64, 839 (1993)28. J. Moore, C. Davis, M. Coplan, Building Scientific Apparatus: a Practical Guide to Design andConstruction (Addison-Wesley, Redwood City, 1989).29. A. DeSilva and G. Goldenbaum, Methods of Experimental Physics, Ch.3 Vol 19 (part A) (1970).30. H. Hokanzono, K. Midorikawa, M. Obara, and T. Fujioka, J. Appl. Phys. Vol. 56, 680 (1984).64


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items