"Science, Faculty of"@en . "Physics and Astronomy, Department of"@en . "DSpace"@en . "UBCV"@en . "Leffler, Steven Roy"@en . "2008-08-29T22:11:08Z"@en . "1992"@en . "Master of Science - MSc"@en . "University of British Columbia"@en . "A Q-switched, cavity-dumped Nd:YAG laser was developed for use as a diagnostic tool in plasma physics experiments. This laser uses an unstable resonator configuration to produce higher gain and better collimation than would otherwise be possible. Pulses of light with energies of 18\u00B12 mJ and FWHMs of about 12 ns were produced. It was found that more than 100 mJ of light was being emitted in the form of prelase pulses, due to the Pockels cells used having a poor extinction ratio. Efforts were made to correct this problem, with only partial success."@en . "https://circle.library.ubc.ca/rest/handle/2429/1585?expand=metadata"@en . "2480152 bytes"@en . "application/pdf"@en . "A Q-SWITCHED, CAVITY-DUMPED YAG LASERFOR USE IN PLASMA PHYSICS EXPERIMENTSbySTEVEN ROY LEFFLERB.Sc. (Honours Physics), The University of Western Ontario, 1990A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF PHYSICSWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIADecember 1992\u00C2\u00A9 Steven Roy Leffler, 1992In 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.(Signature)Department of PhysicsThe University of British ColumbiaVancouver, CanadaDate ^December 18, 1992DE-6 (2/88)AbstractA Q-switched, cavity-dumped Nd:YAG laser was developed for use as a diagnostictool in plasma physics experiments. This laser uses an unstable resonator configuration toproduce higher gain and better collimation than would otherwise be possible. Pulses of lightwith energies of 18\u00C2\u00B12 mJ and FWHMs of about 12 ns were produced.It was found that more than 100 mJ of light was being emitted in the form of prelasepulses, due to the Pockels cells used having a poor extinction ratio. Efforts were made tocorrect this problem, with only partial success.iiTable of Contents Abstract ^ iiList of Figures vAcknowledgements ^ vii1. Introduction 11.1 Motivation ^ 11.2 Summary of Project ^ 21.3 Outline of the Thesis 32. Theory ^ 52.1 Introduction ^ 52.2 The Lasing Medium 52.3 Unstable Resonators ^ 72.3.1 Introduction to Unstable Resonators ^ 72.3.2 Loss Calculation for Unstable Resonators 112.4 Q-Switching ^ 142.4.1 Introduction to Q-S witching ^ 142.4.2 Q-Switched Pulse Characteristics 152.5 Cavity Dumping ^ 17iii3. System Development ^ 203.1 Introduction 203.2 Initial Design ^ 203.2.1 Optical Components ^ 203.2.2 Flashtube Pumping 233.2.3 Timing and Triggering ^ 253.2.4 Expected Performance 283.3 Tests and Design Modifications ^ 303.3.1 Preliminary Tests 303.3.2 Optimization of Output ^ 333.3.3 Attempts to Diagnose and Correct the Pockels Cell Problem ^ 364. Conclusions^ 47List of References 49ivList of Figures Figure 2.1: Simplified energy level diagram for Nd:YAG ^ 6Figure 2.2: Some examples of stable and unstable resonators. (a) Twocommonly-used stable resonator configurations. (b) Two unstableresonator configurations. ^ 9Figure 2.3: Graph of unstable resonator loss due to \"diffraction\", as a functionof the product of the mirrors' curvature parameters. ^ 12Figure 3.1: Schematic diagram of the initial YAG laser cavity layout. ^ 21Figure 3.2: Schematic of the initial Pockels cell triggering circuit design 27Figure 3.3: Variation of cavity-dumped output energy with Pockels cell voltage.Each point is an average of 5 shots. The curve is a weightedleast-squares fit to the data, excluding the four anomalously lowpoints. ^ 34Figure 3.4: Schematic of Pockels cell control circuit with simplified cavitydump system. ^ 38Figure 3.5: Schematic diagram of final YAG laser cavity layout, with Pockelscell control circuit. ^ 40Figure 3.6: Light in the laser cavity (lower trace) and its integral (upper trace).Horizontal scale is 10 ps/division. Vertical scales are arbitrary. ^ 42Figure 3.7: Observed and expected reflection of HeNe light by an apparatusconsisting of a Pockels cell placed between a plane mirror and apolarizer. The theoretical curve is normalized to the same peakintensity as the observed data 45viAcknowledgementsI would like to thank my supervisor, Jochen Meyer, for his constant encouragementand support. I've learned a lot under his guidance. I would also like to thank AbdulElezzabi, Michael Hughes, Ross McKenna, and Peter Zhu for many interesting and usefulconversations about lasers, physics, and life in general. Al Cheuck helped with theelectronics. Dr. Andrew Ng provided the saturable absorbing dye which I experimented within an attempt to improve the laser's performance. Andrew Forsman gave a lot of advice onhow best to prepare and use the dye.I would especially like to thank Hubert Houtman, who laid much of the groundworkfor this project prior to my joining the lab, including fabricating the laser head and some ofthe other components. Hubert also provided a great deal of useful advice on the practicalside of building a laser and getting it to work.Thanks are also due to my family for their support and encouragement, and to thegang at the Lutheran Campus Centre for just being there.The Natural Sciences and Engineering Research Council of Canada supported my firsttwo years of study through their \"Postgraduate Scholarships\" program.viiChapter 1Introduction 1.1 Motivation In most of the plasma physics research done in our lab, experiments are carried outon plasmas generated by ablating material with a 12 J pulse of infrared light from a CO 2 lasersystem. The primary diagnostic tool in these experiments is a ruby laser which produces 6 nslong pulses with energies around 800 mJ. These pulses undergo Thompson scattering offLangmuir waves in the plasma. The scattered light is then analyzed to ascertain the valuesof the parameters of the plasma which are relevant to the experiment being carried out.The principal detector used in these experiments is a Hamamatsu streak camera, whichhas a quantum efficiency which peaks for light whose wavelength is around 420 nm. Therange of wavelengths for which this detector's quantum efficiency is better than 5% extendsfrom about 300 nm to about 530 nm. Since ruby lasers produce light with a wavelength of694.3 nm, the quantum efficiency of the streak camera for ruby laser light is relatively low(around 1.5%). This means that faint scattered signals are often difficult or impossible todetect, and the problem is even worse when the scattering process causes the light to beshifted to a longer wavelength by doppler effects.It was decided that it was not worthwhile to attempt to obtain a streak camera witha higher sensitivity in the red, because this seemed likely to be expensive, if such a camerawere available at all. Instead, the ruby laser would be replaced by a laser which produceslight with a wavelength closer to the middle of the current camera's response curve. Thisthesis describes the construction and testing of a laser to meet this need.1Chapter 1. Introduction^ 21.2 Summary of Project The lasing medium chosen for the new laser was neodymium YAG, which producesnear infrared light with a wavelength of 1.0641 pm. This light can then be frequency-doubled fairly efficiently to produce green light at 532 nm. At this wavelength, the streakcamera's quantum efficiency will be about 5%, so one would expect that for a given amountof input laser power, the response with a frequency-doubled YAG laser would be over threetimes that obtained with a ruby laser, making it easier to measure faint signals.In addition to producing a more useful wavelength of light, YAG lasers produce amuch higher gain for a given amount of input energy. This means that for the laser toproduce the same intensity of light as the old ruby system, it needs a much smaller bank ofcapacitors to produce the pulse of energy which generates the initial inversion. Thissimplifies the design of the laser and reduces its cooling requirements considerably.The initial design goals for the laser were that it was to produce a pulse of light ofapproximately 6 ns duration, with a total energy in the infrared of 50-100 mJ. This wouldbe sufficient to allow amplification to around 1 J via an amplifier to be added later. It wasalso desired that the beam be quite well collimated, because it would have to travel severalmetres in order to reach the experimental target chamber and enter it from an appropriatedirection. An additional requirement was that it had to be possible to accurately synchronizethe firing of the laser with a control pulse, so that the laser could be set up to fire at a fixedtime relative to the firing of the CO 2 laser which generates the plasma being studied.In order to produce pulses of the duration required, it is necessary to both Q-switchand cavity dump the laser. These techniques both involve varying the optical losses in theChapter I. Introduction^ 3cavity with time, so as to cause the laser to produce a brief pulse of light rather than acontinuous beam. The theory behind this is described in Chapter 2.Because of the collimation requirement, and because high-quality mirrors ofappropriate curvature were already on hand in the lab, an unstable resonator configurationwas chosen. As will be discussed in section 2.3, unstable resonators produce very well-collimated beams, as well as having other practical advantages over stable resonators in somecircumstances.In order to allow for synchronization with the existing experimental setup, it wasdecided that an ultraviolet-triggered spark gap should be used to control the timing of thelasing sequence. This allows the laser to be triggered by the same voltage pulse that triggersthe spark gaps in the CO2 laser and its amplifiers, with the relative timing controlled by thelength of delay cable between the last CO2 spark gap and the YAG spark gap.The laser was not frequency-doubled as part of the work described here, but thecrystal for doing so has been obtained and will be added, external to the laser itself, at a laterdate. The technology of frequency doubling will not be discussed further in this thesis.1.3 Outline of the Thesis The remainder of this thesis is organized as follows:In Chapter 2 is found an explanation of some of the laser physics techniques used inthe laser described here, along with formulas for various important parameters that describethe expected performance of the laser. The chapter begins with a brief description of thelasing medium used, Nd 3+ :YAG. It goes on to describe some of the features of unstableresonators, and to explain why they are useful. Formulas describing the loss of light in thisChapter 1 . Introduction 4type of resonator are given, for use later in the thesis. Next, the technique of Q-switchingis described, along with formulas for the energy and duration of pulses produced using thistechnique. Finally, the chapter explains the technique of cavity dumping, which is used toproduce shorter pulses than would otherwise be possible.Chapter 3 explains the initial design for the laser, and makes some estimates of theperformance to be expected from it. The chapter then goes on to describe the tests that wererun on the laser and the modifications to the initial design that were made due to the resultsof these tests. A problem with the Pockels cells used for Q-switching and cavity dumpingwas found, and the chapter concludes with a description of the various attempts that weremade to diagnose and correct this problem. In the end, the cause of the problem was found,but it was not completely solved.In Chapter 4, the results obtained are summarized and suggestions for further workto improve the laser are presented, including suggestions of ways to correct the Pockels cellperformance problems.Chapter 2Theory,2.1 Introduction This chapter contains a summary of those aspects of laser physics which are relevantto the operation of a Q-switched, cavity-dumped Nd:YAG laser with an unstable resonator.It is assumed that the reader is already familiar with the basic structure and principles ofoperation of a simple continuous-wave (CW) laser.The effects of using an unstable resonator, of Q-switching, and of cavity dumping canfor the most part be treated independently, so in this chapter there are separate sections oneach of these techniques, as well as one on the Nd:YAG lasing medium itself.2.2 The Lasing Medium The lasing medium commonly referred to as \"neodymium YAG\" or \"YAG\" iscrystalline, and is composed of yttrium aluminum garnet (Y3A15012), with trivalentneodymium replacing up to 1.5% of the yttrium atoms in the crystal's structure. This isusually written Nd 3+:YAG, or simply Nd:YAG. It was first successfully made to lase byGeusic, Marcos, and Van Uitert [1].Nd:YAG is a very useful lasing material because of its good physical and chemicalproperties, most notably that it is strong and hard enough to be cut and polished withoutsevere breakage problems, and that its high thermal conductivity (0.13 W crri l K- 1 at 300 K)allows for efficient cooling during operation. Also, for a given amount of input energy,Nd:YAG produces a much higher gain than many other solid-state lasing materials. Its index5Chapter 2. Theory^ 6of refraction for wavelengths around 1 pm is 1.82. Koechner [2] summarizes many otherphysical and optical properties of Nd:YAG which are not relevant to the operation of thelaser described in this thesis, and compares Nd:YAG with other common solid-state lasingmedia.At room temperature, the dominant line in the Nd:YAG fluorescence spectrum is at1.0641 pm, and therefore only this frequency will lase under most conditions. Nd:YAG can,however, be made to lase on other lines in its spectrum if a dispersive prism or otherfrequency-selecting device is placed within the laser cavity. The alternative lines with thelowest thresholds for lasing are at 1.0615 pm and 1.0738 pm.All three of the lines mentioned arise from transitions betweenvarious components of the 4F312 and 4I11 ,2 manifolds. A simplifiedenergy-level diagram for Nd:YAG is shown in Figure 2.1, withthe dominant laser transition marked.The process by which YAG lases at 1.0641 pm is asfollows: a powerful lamp is used to provide energy to the YAGrod. Visible-light photons from the lamp are absorbed in therod, exciting electrons from the ground state to the \"pumpbands\", which lie between 18 000 and 25 000 crri l (2.2 -3.1 eV). These electrons rapidly decay to the 4F312 energy level,which has a radiative lifetime of 550 ps. This is called theupper laser level. The laser transition is to the 4Iw2 level (the Figure 2.1: Simplifiedenergy level diagram for\"lower laser level\"), which has a lifetime for decay to the ground Nd:YAG25 ^Pumpbands2015OX 4F3/2':10 100F7.14 1 15/2541 13/24111/20 419/2state of 30 ns.Chapter 2. Theory^ 7In addition to having a fast decay to the ground state, the lower laser level inNd:YAG has a high enough energy to be essentially unpopulated in thermal equilibrium atroom temperature. These qualities make YAG a \"four-level\" lasing medium, so calledbecause there are four distinct energy levels involved in the lasing process. This is veryimportant because it prevents electrons from accumulating in the lower laser level duringoperation, which would reduce the gain of the laser considerably because of the ability ofthese electrons to absorb photons at the laser frequency (which excites the electrons back upto the upper laser level).2.3 Unstable Resonators 2.3.1 Introduction to Unstable ResonatorsAn essential element of all lasers is the resonator: a set of mirrors which confine thelight in the laser so that it can be repetitively amplified by the lasing medium. In most lasers,the resonator is composed of two mirrors, which are either planar or spherical. Generally,the cavity length (or mirror separation) and mirror curvatures are chosen such that theresonator is stable. This means that there exist modes where the light is completely confinedby the mirrors, in the sense that a ray propagating according to geometric optics wouldbounce back and forth between the mirrors indefinitely and never escape past their edges, asillustrated in Figure 2.2 (a). In these resonators, the only mirror-related losses are those dueto transmission through the mirrors and diffraction of light past their edges, both of whichcan be minimized by choosing mirrors of suitable size and quality. A resonator is stable ifit satisfies the relationChapter 2. Theory^ 81^ (2.1)where ,g 1 1 -LIR, is the curvature parameter for mirror i,L is the minor separation,and R. is the radius of curvature of mirror i, defined such that Ri > 0 if the centre ofcurvature of mirror i lies in the direction of the other mirror.One of the disadvantages of stable resonators is that the diameters of the modes oflight which survive in them are almost invariably quite small, because off-axis light tends tobe redirected back towards the centre of the beam, except in neutrally stable resonators suchas the plane-plane type. Yariv [3] solves Maxwell's equations for the case of a beam oflight whose radial intensity distribution is Gaussian, and develops from it a technique forcalculating the form of the lowest-order modes in an optical resonator. This analysisconfirms that the diameter tends to be relatively small (under a few millimetres). The smallmode diameter in stable resonators reduces the gain of the laser, because only a small portionof the volume of the lasing medium can be used for amplification of the light.One way of increasing the active volume of lasing material is to use an unstableresonator, ie. one where the cavity length and mirror curvatures do not satisfy equation (2.1).In these resonators, any ray of light which is not exactly on the central axis of the resonatorwill get farther from it on each bounce, until it eventually misses one of the mirrors andleaves the cavity, as shown in Figure 2.2 (b). This leads to much higher losses than in thecase of a stable resonator, but a much larger beam diameter. If the increase in gain due tothe increased beam diameter is large enough to compensate for the increased diffractionChapter 2. Theory^ 9Figure 2.2: Some examples of stable and unstable resonators. (a) Two commonly-usedstable resonator configurations. (b) Two unstable resonator configurations.Chapter 2. Theory^ 10losses*, using an unstable resonator can be worthwhile.Unstable resonators also have the advantage of providing some control over thetransverse modes. Because light which is propagating off-axis in the resonator is lost morequickly than light which is propagating on-axis, these resonators favour the TEm oo mode overother TEM modes, because it has more of its energy concentrated at the centre of the beam.The TEmoo mode is preferable for many applications because it produces a single-phase,radially-symmetric output beam whose intensity as a function of radial distance is Gaussian.This is in contrast to the higher-order modes, which have a much more complicated structure.A third advantage is that if the mirror curvatures are chosen such that the resonatoris confocal, the output beam will be extremely well collimated [4]. The improvedcollimation occurs because the modes in an unstable resonator are wider than those in a stableresonator, and the limit on beam divergence in the far field imposed by diffraction' scalesaccording to:0a 2.^(2.2)Dwhere 0 is the far-field apex angle of the beam cone,X is the wavelength of the radiation,and D characterizes the diameter of the output beam.Large-diameter beams can thus have a lower divergence in the far-field.* It is conventional to refer to the loss of light past the edges of the mirrors in an unstableresonator as \"diffraction loss\", even though it arises primarily from geometric opticsconsiderations.\" Diffraction in the ordinary sense of the word.Chapter 2. Theory^ 112.3.2 Loss Calculation for Unstable ResonatorsSiegman [5] used geometric optics to produce an approximate expression for theenergy loss due to \"diffraction\" in an unstable resonator. His approach was to assume thatthe light coming from each mirror has the form of a spherical wave of uniform intensity, witha virtual centre which is not necessarily the centre of curvature of the mirror the light iscoming from. By requiring that each virtual centre be the image of the other, on reflectionfrom the appropriate mirror, he was able to derive expressions for the locations of the virtualcentres given the radii of curvature of the mirrors and their separation. From these, he wasable to estimate the amount of energy lost per \"bounce\" by examining how much of eachspherical wave was intercepted by the opposing mirror. This gives the result that the averagesingle-pass fractional intensity loss, gdiff, is given bygdiff 11 - 1 1 - (m2) 4 (2.3) 1 +111 _(g1g2)-1where the upper sign is to be taken when g 1g2 > 1, and the lower sign is to be taken wheng 1g2 < 0. The equation is, of course, not valid for other values of g 1g2, since these correspondto stable resonators according to equation (2.1). The former equation is plotted as a functionof g 1g2 in Figure 2.3.Equation (2.3) is valid for any unstable resonator whose mirrors have sphericalcurvature, regardless of the size or shape of those mirrors, provided that the Fresnel numberof the resonator is large (N a 2ILA, > 1, where a is the radius of the mirrors). The only otherrestriction on the form of the mirrors is that they must both extend past the centreline of theresonator.Chapter 2. Theory^ 12Figure 2.3: Graph of unstable resonator loss due to \"diffraction\", as a function of theproduct of the mirrors' curvature parameters.Chapter 2. Theory^ 13A more detailed treatment of unstable resonator losses, which includes actualdiffraction effects, can be found in Siegman's book [6]. For the purpose of evaluating theperformance of the laser described in this thesis, the simpler analysis which results inequation (2.3) is satisfactory, however.The average loss per pass in a resonator due to sources other than diffraction is givenby:gnd aL -1472 , for gnd < 1^ (2.4)where a is the exponential decay constant for distributed losses in the cavity,and ri is the intensity reflectivity of mirror i.This combines with the diffraction loss from equation (2.3) according to the formula= gnd +gdiff \u00E2\u0080\u0094gndgdiff^ (2.5)to give the total loss for the resonator, g. For most unstable resonators with high-reflectivitymirrors, g gdiff \u00E2\u0080\u00A2A useful parameter for summarizing the losses in a resonator is the photon lifetime,tc. This is the time needed for the energy of the light in the cavity to decrease to lie of itsinitial value, in the absence of any gain mechanism, i.e.dXdt^t(2.6)where r is the energy stored in the cavity.Chapter 2. Theory^ 14The relative rate of decrease in energy due to cavity losses is just the fractional lossper pass, g, divided by the time needed for light to pass through the cavity once:1 dr = cg^(2.7)r dt^nLwhere nL is the optical path length of the cavity,and c is the speed of light.Therefore, the photon lifetime is given byt =nL^(2.8)c cg2.4 0-Switching 2.4.1 Introduction to Q-SwitchingThe technique called Q-switching is often used to allow the production of short,intense pulses from lasers. It works [7] by temporarily increasing the losses in the cavitywhile the lasing medium is being pumped, allowing the inversion of the medium to build upto a level much higher than the threshold where lasing would normally start to occur. Oncepumping is complete, the losses are lowered to their normal value. Because the inversion isthen much higher than the threshold for lasing, light builds up very rapidly in the cavity,resulting in a laser pulse with a very short risetime. After the initial inversion is depleted,the light level in the cavity decays exponentially, with a time constant roughly equal to thephoton lifetime, tc.Chapter 2. Theory^ 15Because light builds up rapidly in the cavity during Q-switched operation, the photondensity is much higher than in ordinary continuous-wave (CW) lasing. This high photondensity supports and is supported by a high level of stimulated emission. The result is thatthe inversion is reduced very quickly to a level far below that at which it started. This meansthat Q-switching can extract energy from the lasing medium much more efficiently than CWlasing can, making it useful any time a high intensity of light is desired.There are many different physical means of achieving the change in cavity lossrequired for Q-switching. The methods which have been used include spinning or vibratingmirrors, electrooptic effects such as the Pockels effect in crystals and the Kerr effect inliquids, acoustooptic effects in crystals, and saturable absorbing dyes. The latter are solutionsof complex organic molecules which have a high absorption coefficient if the intensity oflight is low, but which \"bleach\" and become transparent at high intensities.2.4.2 Q-Switched Pulse CharacteristicsWagner and Lengyel [8] found that by assuming that the change in cavity loss tookplace very quickly, and neglecting slow processes such as pumping and spontaneous emissionduring the buildup of the Q-switched pulse, they were able to describe the buildup of thepulse with a pair of differential equations. In my notation, these are:dy) = cp 1 _ 1di^nt/dn i^2n1=ntdti(2.9)where cp is the total number of photons in the cavity,X2 c t,8nn2tpont Av Vn t =^ (2.11)Chapter 2. Theory^ 16ni is the total inversion in the lasing medium,ti is the time from Q-switching, in units of t,and !it is the total threshold inversion.The first of these equations expresses the fact that each stimulated emission increasesthe number of photons in the cavity by one, and that on average one photon leaves the cavityevery t seconds due to the cavity loss, g. The second equation expresses the fact that foreach photon created by stimulated emission, the total inversion of the lasing mediumdecreases by two. This is because an electron moves from the upper laser level to the lowerlaser level, and the lifetime of the lower laser level is long compared to the risetime of theQ-switched pulse, even in a four-level laser.Wagner and Lengyel were able to solve this system of equations and find anexpression for the amount of energy released by stimulated emission:n.hvX\u00E2\u0080\u0094 t2 , for nt >nt (2.10)where ni is the total initial inversion in the lasing medium,h is Planck's constant,and v is the frequency of the radiation.The total threshold inversion is given by the formulawhere n^is the index of refraction of the medium,t spont is the radiative lifetime of the lasing medium,Chapter 2. Theory^ 17Av^is the width of the spontaneous fluorescence spectrum,and V^is the volume of lasing material present.The fact that the energy given by equation (2.10) does not depend on the cavity lossesis due to the assumption that the threshold inversion can be ignored relative to the initialinversion, since under these conditions the Q-switched pulse will build up rapidly, and so theeffect of cavity losses can be ignored during the time when energy is being released bystimulated emission.Koechner [2] uses Wagner and Lengyel's approach to produce an expression for theduration of the Q-switched pulse:n i - nf0t t ^ (2.12)+1n(n i ln,)]where of is the total inversion left after lasing is complete.For large ni, Atp will be only slightly longer than te .2.5 Cavity DumpingThe technique of cavity dumping is a way of getting energy out of a laser moreefficiently than the usual method, which is to make one of the end mirrors partiallytransmitting so that light leaks through to be used as the laser's output. In a cavity-dumpedlaser, both mirrors are as close to 100% reflective as possible, and an electrooptic oracoustooptic device is used to \"switch\" the light out of the cavity when its intensity reachesa maximum. This technique can be used alone on a CW laser, or can be combined with Q-switching to produce shorter, more symmetrical, pulses than would be produced by the latterp^cChapter 2. Theory^ 18technique alone. The combination of cavity dumping and Q-switching is frequently referredto as \"pulse transmission mode\" (PTM) Q-switching [9].If the device used to switch light out of the cavity changes state in a time muchshorter than the cavity round-trip time and introduces a loss near 100% to the cavity, thenthe width of the resulting output pulse will be close to the round-trip time of the cavity. Thisoccurs because in one round-trip time, all the light which was in the cavity prior to the timeof switching passes the switching device and is deflected out of the cavity. If the laser isboth Q-switched and cavity-dumped (PTM), the energy of the output pulse will beapproximately the same as that which would have been obtained from an equivalent laserwhich is only Q-switched [10]. Its energy will thus be given by equation (2.10).The reason why a PTM laser produces a pulse with the same amount of energy as anequivalent Q-switched laser's is that when either laser's Q-switch triggers, light energy buildsup very rapidly in the cavity, so that much of the stimulated emission takes place early in theQ-switched pulse's development. In the non-cavity-dumped case, this energy then decaysrelatively slowly by transmission through the output mirror. During this stage, there is stilla small amount of stimulated emission occurring, but it does not add a significant amount ofenergy to the light in the cavity. In the PTM case, the peak energy in the cavity is rapidlyremoved by the switching device, producing a pulse which is shorter, but which does nothave significantly less energy.Since the energies in the two types of pulse are essentially the same, while the cavity-dumped pulse's length is generally several times shorter (assuming a reasonable outputcoupling loss for the non-dumped case), it is clear that the intensity of the light produced byChapter 2. Theory^ 19a PTM system is generally higher than that from an equivalent Q-switched system withoutcavity dumping.The next chapter describes the design and development of a laser system whichcombines the short pulse length and high intensity of a pulse transmission mode system withthe high gain and small beam divergence of an unstable-resonator YAG laser. The expectedperformance of the laser is estimated using the formulas from the present chapter.Chapter 3System Development3.1 Introduction The design of the Nd:YAG laser system was changed several times over the courseof this thesis project, to correct problems that were discovered and to improve performance.This chapter describes the initial design and performance estimates for the laser, the tests thatwere run on it, and the design changes that were made as a result of those tests.3.2 Initial Design 3.2.1 Optical ComponentsThe initial design for the Nd:YAG laser described in this thesis was based on thedesign of the ruby laser it is to replace [11]. It used two Pockels cell/polarizer pairs tointroduce controlled losses to the cavity: one for Q-switching, and one for cavity dumping.A schematic of the original layout of the YAG laser is shown in Figure 3.1.As shown, the resonator is composed of two mirrors, labelled M 1 and M2. M1 is2.2 cm in diameter and is concave with a radius of curvature of 10 m. M2 is the same size,but is convex with a radius of curvature of 7 m. Both mirrors are high-reflectivity dielectriclaser mirrors. The initial perpendicular separation of the mirrors was 1.58 m, which was20Chapter 3. System Development21Figure 3.1: Schematic diagram of the initial YAG laser cavity layout.Chapter 3. System Development 22chosen to make the effective length of the cavity for purposes of geometric ray tracing* equalto 1.5 m, so that the mirrors would be confocal. The round-trip time of light in the cavitywas 11.6 ns.The Nd:YAG laser rod is circular in cross-section, with a diameter of 6.35 mm anda length of 10 cm. The ends of the rod are antireflection coated (1.064 pm) and are parallel,with a wedge angle of 6\u00C2\u00B0 from normal. The doping level is high, with 1.1% of the yttriumatoms in the crystal having been displaced by neodymium atoms. For efficient pumping ofthe rod, it is mounted in a metal reflector whose cross-section has the form of twooverlapping ellipses, with one focus of each coinciding. The rod is located at that focus, anda xenon flashtube is located at each of the other two foci, so that the light from the flashtubesis focused into the rod. The rod is cooled by a continuous flow of room-temperature waterover its surface, and the flashtubes are cooled by air flowing through their mounting tubes.The Pockels cells, PC 1 and PC2, are both 9 x 9 x 25 mm lithium niobate (LiNbO 3)crystals. Each is cut such that the optical axis lies along the length of the crystal. The endfaces have an anti-reflective coating, which is optimized for 1.064 pm. One pair of opposingfaces perpendicular to the beam path have gold electrodes deposited on them. The theoreticalDC quarter-wave voltage for these crystals is 1 514 V at 1.064 pm.The polarizers, P 1 and P2, are of the standard Glan-Foucault type, and are orientedsuch that the horizontal polarization is transmitted and the vertical polarization is deflected* The effective length referred to here is not the optical path length, which is of courselonger than the physical length of the cavity. This distinction should be familiar to thereader: a stick appears to get shorter when it is stuck into a bucket of water, despite thefact that the optical path length to its end has increased. (I am indebted to H. Houtmanfor this analogy.)Chapter 3. System Development^ 23out of the cavity. Because their faces are not antireflection-coated, the polarizers were angledslightly to prevent the formation of sub-cavities.In addition to the above components, an iris, labelled \"I\" in the diagram, wasintroduced to the cavity to limit the diameter of the propagating modes of light to less thanthe diameter of the laser rod, so as to prevent light from scattering off of its unpolished sides.This is formally equivalent to reducing the sizes of both end mirrors, and so to first order itintroduces no additional loss to the cavity, as discussed in section 2.3.2. With the iris inplace, however, essentially all of the unstable resonator loss is \"due\" to light being interceptedby it. The amount of light intercepted is sufficiently large that the intensity in the cavity wasmonitored by tilting the iris at a slight angle and attaching an annular aluminum reflector toit, to deflect light towards a nearby high-speed photodiode. This was omitted for clarity inthe figure above, but is shown in Figure 3.5.Because the mirrors M 1 and M2 are partially transparent to visible light, the laser couldbe aligned by running the beam from the helium-neon (HeNe) laser through the back of M 1and down the length of the cavity with the aid of plane mirrors M3 and M4. The reflectionfrom M2 was large enough that the reflected pulse could be traced back to P 1 to ensure thatthe Pockels cells and mirrors were oriented normal to the beam. A removable aperture wasplaced between P1 and the YAG rod during alignment to aid in detection of the reflected spot.3.2.2 Flashtube PumpingAs already mentioned, the initial inversion in the YAG rod is produced by a flash oflight from two xenon flashtubes. The tubes chosen have an arc length of 3\", a bore diameterof 9 mm, and a rated maximum energy per flash of 600 J. Their operating voltage range isfrom 1.5 kV to 2.5 kV.Chapter 3. System Development 24Flashtube lifetime is characterized by the ratio of the energy of the applied voltagepulse to the tube's \"ultimate limit\": the maximum energy the tube can handle withoutshattering [12]. The ultimate limit is given, in Joules, by the empirical formulaUilm = 90 LD/ (3.1)where L is the arc length, in inches,D is the bore diameter of the tube, in mm,and T is the flash duration, in ms.The units used for L and D in this equation were chosen to match the units used byEG&G in the specifications for the flashtubes they manufacture. It seems odd, but their tubeshave bore diameters which are integral in millimetres, and arc lengths which are integral orhalf-integral in inches. I presume that this is due to a preference for specifications in inchesbeing frustrated by quartz tubing only being available in metric diameters.A flash duration of 43.5 ps was chosen to make the pumping time much less than thelifetime of the upper laser level, so that there would be little loss due to spontaneousfluorescence and nonradiative decays. Substituting this into the equation gives U li. = 507 J.The manufacturer's recommendation for a flashtube enclosed in a reflective cavity, as is thecase here, is for the flash energy not to exceed 40% of Ufifri, because the large amount of lightreturned to the tube by the reflector increases the tube's thermal stress.The energy for the flash comes from two two-stage pulse-forming networks, whichhave characteristic impedances of about 0.3 S2\u00E2\u0080\u0094to roughly match the resistance of theflashtube arcs\u00E2\u0080\u0094and total capacitances of about 71.6 pF. These produce simultaneous 43.5 psChapter 3. System Development^ 25long current pulses in each tube, with an energy given by 0.5 CV', where V is the initialcharging voltage of the networks.The voltage on the networks is limited at the low end by the minimum operatingvoltage of the flashtubes, and at the high end by the thermal stress considerations discussedabove. This results in an operating voltage range of 1.5 kV to 2.38 kV, or an energy rangeof 80 J to 203 J per pulse. To minimize tube wear, and to allow room to compensate fordeterioration of the flashtubes with age, the charging voltage used for most firings of the laserwas in the neighbourhood of 1.74 kV, which corresponds to a pump energy of about 110 J.According to reference 12, this should result in a tube lifetime of better than 1 000 shots.3.2.3 Timing and TriggeringOnce the pulse-forming networks are charged, the laser is triggered by a push buttonconnected to a two-channel electronic delay unit, located in an electromagnetically-screenedroom. This unit produces two 40 V output pulses at times which can be set independentlyin intervals of 100 ns. The first pulse, at t = 0, goes to a silicon-controlled rectifier unitwhich produces a short 400 V pulse. This pulse is stepped up to 20 kV by a transformer, andis applied to the metal reflector surrounding the YAG rod and the flashtubes. The highelectric field caused by this pulse makes the gas in the flashtubes break down, and currentis supplied to them by the pulse-forming networks described above.Near the end of the flashtube pulse, the second channel of the delay unit triggers,sending a 40 V pulse to a krytron unit which produces a 23 ns long, 5 kV output pulse,which is used to trigger the ultraviolet-triggered spark gap which controls the two Pockelscells. When the laser system is used as a diagnostic with the 12 J CO 2 laser, this krytron unitChapter 3. System Development^ 26will no longer be necessary, as the spark gap will be triggered directly by the trigger pulsethat comes out of the CO2 laser's spark gaps.A schematic diagram of the initial design for the Pockels cell triggering circuitry isshown in Figure 3.2. This design was based on that of the circuit used in the existing rubylaser. Until the gap is triggered, coaxial cables L 1 and L2 are held at the quarter-wave voltageof the cell, V114, which means that PC 2 is at V114, while PC 1 is at 0 V. The cavity is thus ina high-loss situation, because of the voltage on PC 2. When the spark gap is triggered by theapplication of a high-voltage pulse to its trigger pins at time tg, it becomes conductive,allowing the voltage in the charged cables to flow through into cables L3 and L4. Adecreasing step pulse propagates back up cables L 1 and L2 towards their ends. This pulsereaches PC2 at time tg + 3 /1/(2c), where / 1 is the length of cable L1 . This causes the voltageacross PC2 to drop rapidly to 0 V, which means that the loss in the cavity should drop to itsdiffraction-limited value, allowing light to begin to build up via stimulated emission. At timetg 313/(2c), the voltage pulse in cable L3 reaches PC1 , causing the voltage across it to rapidlyrise to V114 , which causes the light in the cavity to be dumped out of polarizer P 1 . The lengthof cable L3 is chosen such that this occurs when the intensity of the light in the cavity reachesits peak.The length of time for which the cavity-dumping pulse is applied to PC 1 is controlledby the length of cable L2. Specifically, the duration of the dump signal will be given by3 /2/c . This was initially chosen to equal 6 ns, the desired output pulse length.The 4 CI resistor and 10 nF capacitor wired in parallel to the gap help maintain thedischarge and ensure that the initial current is high\u00E2\u0080\u0094over 300 A. This decreases theChapter 3. System Development27Figure 3.2: Schematic of the initial Pockels cell triggering circuit design.Chapter 3. System Development^ 28resistance of the gap, allowing the voltage step pulses to pass through with less distortion oftheir shape.3.2.4 Expected PerformanceThe expected performance of the laser can be estimated by using the formulas givenin Chapter 2 and the component specifications given in section 3.2.1. While the estimatesobtainable will not be extremely precise due to the need to make some assumptions along theway, they will illustrate the general performance to be expected from the laser.Substituting the appropriate values into equation (2.3) shows that the diffraction lossin the YAG laser's cavity should be approximately 0.29 per pass. The loss due totransmission through the mirrors will be negligible, so the only other losses are those due toscattering off of components (particularly the polarizers, which are not antireflection-coated),and due to absorption in the YAG rod and other parts. All of these losses can be includedin the distributed loss constant, a, in equation (2.4). A reasonable approximation for theseis to take gni\u00E2\u0080\u0094 0.05. Substituting and gdiff into equation (2.5) gives g ----- 0.33.Putting the above loss per pass into equation (2.8), the photon lifetime in the cavityis calculated to be about 18 ns.The total threshold inversion of the laser is given by equation (2.11). The index ofrefraction, n, of Nd:YAG is 1.82 in the vicinity of 1 pm. The width of the spontaneousfluorescence spectrum of YAG, Av, is about 1.8x10 11 Hz. Substituting these values into theequation gives a threshold inversion of 4.4x10 15.In order to calculate the pulse energy and other parameters, the initial inversion of thelaser is also needed. Because the duration of the flashtube pulse is much less than theChapter 3. System Development^ 29spontaneous fluorescence lifetime of the YAG rod's upper laser level, the total initialinversion can be estimated by using the formulaU lamp^ (3.2)hv/aserwhere U/amp is the energy of the electrical pulse applied to the flashtubes,and e^characterizes the efficiency of energy transfer from the electrical pulse throughto the electrons in the upper laser level.Following Yariv [3], the efficiency with which the flashtubes convert electrical energyinto light can be roughly estimated to be about 0.5. About 5% of the light will fall withinthe absorption bands of the YAG crystal, and only about 5% of that will actually beabsorbed. Finally, when an electron decays from the absorption bands to the upper laser levelit loses energy, which must be accounted for within e. The fraction of energy remaining afterthis decay will be roughly equal to v iaser iv iamp \u00E2\u0080\u0094 0.5. Combining these factors gives a valuefor e of about 6.25x10 -4 .Assuming a lamp energy of 110 J and the above value for e, equation (3.2) givesn i p 3.7x10 17 . The ratio of initial to threshold inversion is thus^85 > 1, thereforeequation (2.10) can be used to estimate the energy of the Q-switched and cavity-dumpedpulse. Substituting the above results into the equation gives a pulse energy of about 35 mJ.While this is a bit less than the goal for this laser, it is not a major problem, as it wasexpected that an amplification stage would need to be added before the laser could be usedto replace the existing ruby laser (which also used an amplifier to boost its output energy).Chapter 3. System Development^ 30Because ni > n\u00E2\u0080\u009E it follows that n,:\u00E2\u0096\u00BAnf , so equation (2.12) can be used to calculate thewidth of the Q-switched pulse in the laser when cavity dumping is disabled. The resultingwidth is about 19 ns, only a little longer than tc .3.3 Tests and Design Modifications 3.3.1 Preliminary TestsThe first test made during construction of the laser was that the flashtube pulse wasof the correct length. This was determined by measuring the current passing through thetubes by means of a Rogowski coil wrapped around their ground leads. The full width athalf-maximum (FWHM) of the current pulse was 50\u00C2\u00B14 ps, which is reasonably close to theexpected value of 43.5 ps. A measurement taken much later, after the tubes had been firedhundreds of times, gave a pulse width of 44\u00C2\u00B14 ps, with a slightly less square pulse shape.This change in the shape and duration of the pulse may be due to erosion of the flashtubeelectrodes, although no other deterioration in performance has been noted.A great deal of time was spent constructing and adjusting the spark gap to control thePockels cell timing. The design chosen, like that of the timing circuit as a whole, was basedon that used in the existing ruby laser. This proved to be less than ideal, as the ruby laserused KDP * Pockels cells, which have a much higher quarter-wave voltage than the LiNbO 3cells used with the YAG laser. Because the voltage being switched was many times smaller,the electrodes in the spark gap had to be much closer together in order for them to breakdown properly. This caused problems because the thread of the screws used to adjust the* KDP is an abbreviation for \"potassium dihydrogen phosphate\", KH2PO4.Chapter 3. System Development 31electrode separation was too coarse to allow accurate adjustment of such a small gap. Thesmall electrode separation also caused great problems in triggering the gap, because the smallgap and relatively large electrode diameter made it difficult to get enough ultraviolet lightfrom the trigger pins into the gap to break it down.Because of these problems, it took a fair bit of effort to get the gap adjusted so thatit would trigger reliably, and it required readjustment more frequently than would be desiredin a working laser system. No attempt was made to accurately measure the jitter of the gap,because it was apparent that a new spark gap or other suitable triggering device would haveto be constructed before the laser could be incorporated into the experimental system. Jitterin the spark gap's triggering was not a problem for purposes of testing the laser, as the timingcircuitry always maintains a fixed delay between Q-switching and cavity-dumping, regardlessof spark gap jitter. The length of the delay between the flashtube pulse and the triggeringof the Q-switch only needs to be accurate to within a few microseconds, so spark gap jittercauses no problem there either.Once the spark gap was triggering reliably enough to run tests, its performance wasstudied by connecting cable L3, which would normally go to the cavity-dumping Pockels cell,to an oscilloscope via a series of attenuators. The initial results were far from the ideal1514 V square pulse. This problem turned out to be due to the connectors used to join thecables to the Pockels cells not being able to handle the voltage needed, and due to attenuationin the cables. Replacing the wiring with thicker coaxial cable and using high-voltageconnectors solved both problems, resulting in a 7 ns FWHM pulse with the correct amplitude.The shape of the pulse was still a problem, however, as it had a risetime of 4 ns, and asimilar decay time.Chapter 3. System Development 32Before any attempt was made to solve the cavity dumping voltage pulse shapeproblem, the laser was tested out with cavity dumping disabled. The light intensity in thecavity was monitored by means of a PIN photodiode aimed at the intra-cavity iris, and Q-switched pulses with FWHMs of 30\u00C2\u00B14 ns were observed to occur 60\u00C2\u00B14 ns after the spark gapwas triggered. The pump energy was varied from 106 J to 155 J and no significant changein the Q-switched pulse build-up time was observed.The variation in the height of the Q-switched pulse was also studied as a function ofthe delay between the firing of the flashtubes and the triggering of the spark gap. Asexpected, if the delay was less than about 40 ps the intensity of the Q-switched pulse wasgreatly reduced, because the gap was then firing before the flashtubes had finished pumpingthe YAG rod. For delays between 40 ps and 80 ps, the intensity was relatively insensitiveto the exact length of the delay, because these times are much shorter than the lifetime of theupper laser level. The delay setting chosen for permanent use was 75 ps.It was decided that because of the relatively long risetime of the cavity dumpingvoltage pulse, the length of the pulse should be increased from a nominal value of 6 ns to12 ns, in order to have a longer period of constant voltage on the Pockels cell. A secondreason for this change was that it was realized that the cavity round-trip time was fixed at11.6 ns by the requirement that the mirrors be confocal, and having a cavity-dumping timethat is less than the cavity's round-trip time would be inefficient, as only part of the energyin the cavity would be dumped.Finally, cavity dumping was enabled, and output pulses with a FWHM of 12\u00C2\u00B14 nswere observed. There was a significant variation in the height of the pulses from shot toshot, which was attributed to the lack of longitudinal mode control in the laser cavity. In theChapter 3. System Development 33existing ruby laser, this problem was solved by the introduction of a Fabry-Perot etalon. Thissolution cannot be applied to the YAG laser because it does not use planar mirrors, so theetalon would distort the curvature of the modes in the cavity, ruining the collimation of thelaser's output.The energies of the output pulses were measured with a GenTec Joule meter, and werefound to vary from 0.15 mJ to 3.6 mJ\u00E2\u0080\u0094far below what was expected.3.3.2 Optimization of OutputBy means of various improvements to the alignment of the cavity, the energy rangeof the output pulses was increased to 6 - 10 mJ. The light in the cavity was monitored withand without cavity dump, and there was not much difference between the two. It was clearthat no more than 10 - 15% of the light in the cavity was being dumped. The timing of thedump signal was checked and appeared to be fine.The energy output was studied as the Pockels cell voltage was varied, and the resultsare shown in Figure 3.3. Despite the fact that an average of several shots was used at eachvalue of voltage, there are four anomalously low points on the graph. These seem torepresent temporary reductions in the laser's output, and may be due to thermal lensing andbirefringence in the rod due to a buildup of heat when many shots were taken in succession.Their occurrence was not generally repeatable. The curve shown on the graph is a weightedleast-squares fit to the data, excluding the four anomalous points. The peak energy outputseemed to occur for Pockels cell voltages between 1.8 and 1.9 kV, which is significantlyhigher than the theoretical value, 1.514 kV.Measurements using the helium-neon alignment laser showed that the DC quarter-wave voltage for the HeNe's 632.8 nm beam was also higher than its theoretical value, andChapter 3. System Development^ 34Figure 3.3: Variation of cavity-dumped output energy with Pockels cell voltage. Eachpoint is an average of 5 shots. The curve is a weighted least-squares fit to the data,excluding the four anomalously low points.Chapter 3 . System Development^ 35by the same amount, once the effects of the frequency difference were taken .into account.This effect seems to represent some problem with the crystal itself.The Pockels cell voltage setting chosen for future use was 1.822 kV, as this seemedmore or less optimum. Generally pulses in the 10 - 11 mJ range were produced. Withcertain mirror alignments, pulses as large as 16 mJ could be obtained, but the shot-to-shotenergy stability was drastically reduced.The variation in the laser's output energy was also studied as the pump energy wasvaried from 99 J to 180 J, and no significant increase was observed. This result wassomewhat unexpected, as the amount of energy released should increase with greater pumpenergy, unless the inversion in the rod is saturated. Based on the estimates of section 3.2.4,pump energies in the range studied should only be sufficient to excite electrons to the upperlaser level in 1% - 2% of the Nd atoms in the crystal.In the course of studying this strange behaviour, it was discovered that when the laserwas fired, 130 - 170 mJ of energy came out of the Q-switch polarizer. When cavity dumpingwas disabled, this energy increased to 175 - 185 mJ. Photodiode measurements showed astrong pulse of the same general shape as the previously-observed intracavity light, comingout of the polarizer at about the same time as light builds up in the cavity. This indicatedthat the Pockels cell/polarizer combination was not producing zero loss after triggering,contrary to expectations, and also suggested that the cavity-dumping cell and polarizer werenot introducing 100% loss, because otherwise a greater fraction of the light in the cavitywould be coming out of the cavity-dump polarizer.Chapter 3. System Development 36As this problem with the Pockels cells was clearly causing the loss of the majority ofthe energy produced by the laser, efforts from this point on focused on finding the cause ofthis problem and fixing it. These efforts are discussed in the next section.3.3.3 Attempts to Diagnose and Correct the Pockels Cell ProblemThe first test performed on the Q-switch cell was to run the beam from the helium-neon laser through it and see how much light was deflected out of the cavity as a functionof Pockels cell voltage. Because of the geometry of the cavity, the cell was between parallelpolarizers, as shown in Figure 3.1. This test indicated that the cell had an extinction ratio*of 50\u00C2\u00B120, which seems rather low. Koechner [2] mentions that an ordinary extinction ratiofor a Pockels cell in a laser system is \"a few hundred\".The effect on extinction of the pressure of the clamps holding the LiNbO 3 crystals inplace was investigated, because Koechner states that clamping can introduce somebirefringence in these crystals. It was found that reducing the clamp pressure did improvethe extinction ratio slightly, but this caused no improvement in the laser's performance.The alignment of the polarizers was also improved, to make certain that their axes ofpolarization were accurately parallel to one another. This improved the extinction ratio to230\u00C2\u00B1110, but caused no real improvement in the laser's performance.Next, the voltage pulses being applied to the Pockels cells were examined. Attemptswere made to measure the pulses by various direct and indirect means, but no way of doingso was found which would allow measurement on a fast enough timescale to be useful* The extinction ratio for a Pockels cell is normally determined by placing the cellbetween crossed polarizers, and finding the ratio of the maximum to the minimumtransmitted light intensity as the voltage is varied from zero to V,. This is slightlydifferent from the test conditions described above, but the results should be equivalent.Chapter 3. System Development 37without altering the shape of the voltage pulses being measured. Attention was then focusedon improving the shape of the cavity-dumping electrical signal, as viewed by connecting thecable that normally goes to the cavity-dumping cell to a scope via an attenuator.As already mentioned, the shape of this pulse was not particularly square. It wasimproved by simplifying the switching circuitry so that the electrical signal used for cavitydumping came directly from the charged cable connected to the Q-switch cell, as shown inFigure 3.4. This simplified the circuit and improved the impedance matching at the sparkgap. The resulting electrical signal was a very nice square pulse with an amplitude of880\u00C2\u00B150 V, a FWHM of 12.2\u00C2\u00B10.4 ns, and a risetime of 1.6\u00C2\u00B10.4 ns, indicating that the voltageon the Q-switch cell had to be dropping all the way to zero in a similar amount of time. Thisresulted in a slight improvement in the laser's performance. In combination with animprovement in the positioning of the cavity-dump polarizer, the above change increased thetypical output energy of the laser to 18\u00C2\u00B12 mJ. On occasion, pulses with energies as high as23 mJ were observed. There was still approximately 120 mJ coming out of the Q-switchpolarizer on each shot, however, indicating that the pulse shape had not been the root causeof the problem.Another suspected cause was that there might be some sort of self-focusing effectoccurring in the Q-switch cell due to the beam having a relatively narrow diameter there (itbeing closer to the convex mirror than the other cell, as shown in Figure 3.1). To see if thiswas the cause of the problem, the wiring to the two Pockels cells was reversed, so that thecell which had formerly been the cavity-dump cell now became the Q-switch cell, and viceversa. No significant improvement in performance was observed, and lasing becamesomewhat erratic, so this approach was abandoned.Chapter 3. System Development38Figure 3.4: Schematic of Pockels cell control circuit with simplified cavity dump system.Chapter 3. System Development 39Some problems with poor Q-switch performance using KD*P t Pockels cells havebeen reported to be caused by fast switching of the voltage on the cell inducing an acousticwave in it [13]. This interferes with proper Q-switching because the electrooptic coeffi-cients change locally when there is stress on part of the crystal. An attempt was made to seeif increasing the risetime of the pulses on the YAG's Pockels cells had any effect by wiringcapacitors of various sizes across them, to increase the risetime and decay time of the voltagepulses. This made the laser very unreliable and produced no improvement in output energy.Finally, it was decided to simplify the laser by using a single Pockels cell andpolarizer to both Q-switch and cavity dump the system. A schematic of the resulting layoutand triggering circuitry is shown in Figure 3.5. This change simplifies alignment of thecavity and reduces losses, because the laser contains fewer optical components. Because ofthe removal of components, the mirror separation was decreased to 1.56 m, which resultedin a round-trip time for light in the cavity of 11.2 ns. These changes do not significantlyalter the results of section 3.2.4.The triggering circuit used with this configuration is the same as that shown inFigure 3.4, except that the cable which formerly went to Pockels cell 1 is now leftunterminated, so that the voltage pulse which formerly went to PC 1 is now reflected back toPC2. The lengths of the cables were adjusted to ensure that this pulse would arrive at PC 2at the appropriate time for cavity dumping.No significant change in the performance of the laser was observed as a result of theremoval of the Pockels cell, but the change did result in a major decrease in the amount ofelectrical noise being picked up by the PIN photodiodes used to measure the light in theKD *P is the standard abbreviation for deuterated potassium dihydrogen phosphate.Chapter 3. System Development ^40 a)zx Figure 3.5: Schematic diagram of final YAG laser cavity layout, with Pockels cell controlcircuit.Chapter 3. System Development 41cavity and the light coming out of the polarizer, since there was now less high-voltage wiringnear them. This made visible something which had not been seen previously: a series of low-intensity pulses of light in the laser beginning about 35 ps before the Pockels cell wastriggered. When the light in these prelasing pulses was integrated with a simple RC circuit,it was found that they were taking up the majority of the energy from the laser rod, as shownin Figure 3.6.Some notes on the figure are in order. The traces begin at the firing of the flashtubes.The spike which extends off-scale about 75 ps later is the Q-switched light in the cavity. Itis an order of magnitude larger in intensity than the prelasing pulses. The integrated traceshows that the energy in the Q-switched pulse was no more than 20% of the total energyextracted from the YAG rod. This figure is comparable to the ratio of the difference inenergy coming out of the Q-switch polarizer with and without cavity dumping to the energythat comes out without dumping (see p. 35). It should be noted that while the RC time ofthe integrator used is clearly not quite long enough for accurate integration, the traceproduced still provides an upper bound on the ratio of the energy of the Q-switched pulse tothe total energy extracted, which is sufficient for diagnostic purposes.Clearly, what was happening in the laser was that the Pockels cell, even when heldat its quarter-wave voltage, was not introducing sufficient loss to the cavity to hold off lasing.As a test, the beam path was blocked by placing a piece of white cardboard in the cavity.If the cardboard was placed anywhere but between the concave mirror and the Pockels cell,no lasing took place. If it was placed in the latter location, a small amount of prelasing wasobserved, beginning about 20 ps before the Q-switch was triggered. This indicates that eventhe small amount of reflection from the Pockels cell's antireflection-coated end faces isChapter 3. System Development^ 42Figure 3.6: Light in the laser cavity (lower trace) and its inte-gral (upper trace). Horizontal scale is 10 ps/division. Verticalscales are arbitrary.Chapter 3. System Development 43sufficient to cause prelasing, due to the high gain of the YAG rod. This prelasing, however,was negligible compared to that without the cardboard, indicating that the main problem wasthat the Pockels cell was not producing enough loss, rather than that the reflection from itsfaces was too large.This problem with prelasing is reported by Koechner [2] to be extremely commonwith YAG lasers, due to the high gain of the Nd:YAG lasing medium.An attempt was made to decrease the effect of this prelasing by decreasing the lengthof the delay between the flashtubes firing and the Q-switch being triggered. This was notfound to be useful, as decreasing the delay led to postlasing because the Pockels cell wasbeing switched before the flashtubes had finished pumping the rod, so that a lot of energywas left in the cavity after the Q-switched pulse was produced.It was suspected that the cause of the poor Pockels cell performance might be that theelectric field in it was not sufficiently uniform, due to the fact that the electrodes on thecrystal's holder were about the same size as the crystal, which could lead to a nonuniformfield due to edge effects. To solve this problem, a new Pockels cell holder was made whichhad larger electrodes. This did not improve performance, but it was found that decreasingthe pressure on the crystal by loosening the new mount reduced the energy in the prelasingpulses by about one half, when the mount was as loose as was possible. The limiting factorin how loosely the crystal could be clamped was that the weight of the high-voltage cableconnected to the mount put stress on the upper electrode which would cause the position ofthe crystal to shift if the mount was too loose. An effort was made to reduce this stress inthe redesign of the electrode, but this was only partially successful.Chapter 3. System Development 44In a final attempt to solve the prelasing problem, a cell filled with a saturableabsorbing dye* was added to the cavity. Such dyes are often used on their own to producepassive Q-switching when accurate timing is not needed. They suffer from the drawback thatthe Q-switched pulses produced have a significant \"jitter\" in the time at which they areproduced. It was hoped that by using a weak solution of the dye in addition to the Pockelscell Q-switch, enhanced cavity loss prior to Q-switching could be achieved withoutintroducing extra jitter in the timing, because the Q-switch time would be controlled by thePockels cell.When dye of the appropriate concentration was placed in the cavity, some shots withno prelasing were observed. Unfortunately, after only a few shots the Q-switched light pulsebegan coming later and later, until eventually timing jitters as large as several microsecondswere observed. Allowing the dye to \"rest\" for ten to fifteen minutes sometimes made itpossible to fire one or two good shots before the jitter reappeared, but this limitation wasunacceptable, and so the use of the saturable absorbing dye was abandoned.As a final test, the extinction ratio of the Pockels cell was measured more accuratelythan had been done previously. The cell was placed between a plane mirror and a polarizer,and the intensity of helium-neon laser light reflected by this apparatus was measured. Thisapproach reduces the possibility of error due to imperfect polarizer orientation. Interferencebetween the HeNe beam and its reflections was also eliminated, to improve the stability ofthe output. The results are plotted in Figure 3.7.* The dye used was Bis (4-dimethylaminodithiobenzil) nickel, dissolved in 1,2-dichloro-ethane. No attempt was made to determine the exact concentration used, because theconcentration needed was expected to change with cavity alignment. The dye wasprepared by diluting a saturated stock solution to about 1/25 of its original concentration,this ratio having been found by trial and error to be about ideal.400^600^800^1000Pockels Cell Voltage (V)Chapter 3. System Development^ 45Figure 3.7: Observed and expected reflection of HeNe light by an apparatus consisting ofa Pockels cell placed between a plane mirror and a polarizer. The theoretical curve isnormalized to the same peak intensity as the observed data.Chapter 3. System Development 46As shown, the output intensity was observed to peak at about 100 V, rather than at0 V. This shift could be due to a static birefringence in the crystal, but introducing such aneffect to the theoretical plot does not improve the overall fit to the data. The position of theminimum was within 40 V of the expected value in these measurements, but this position wasobserved to vary from time to time. Positions of the minimum as much as 150 V higherwere observed. The extinction ratio of the data plotted is 31\u00C2\u00B14, which agrees reasonably wellwith the less accurate measurements reported on page 36.The cause of the problem in the laser's performance has thus been identified, allowingthe formulation of some possible solutions. These are discussed in the next chapter, whichalso contains a summary of the results reported in this thesis, and some other suggestions forimproving the laser's performance.Chapter 4ConclusionsA Q-switched, cavity-dumped Nd:YAG laser system has been developed which iscapable of producing 18\u00C2\u00B12 mJ pulses with a full width at half-maximum of about 12 ns.While the pulse energy observed is more than half of the initial estimate of what the outputenergy should be, it is apparent that this estimate was too low, as there is over 100 mJ ofenergy being lost in prelasing which could in principle be recovered as useful output. Thecause of this prelasing is that the Pockels cell used does not introduce enough loss to thecavity to counteract the high gain of the YAG rod.Despite numerous attempts to improve the performance of the Pockels cell, itremained unsatisfactory. The loss introduced by the cell was measured as a function ofapplied voltage, and it was found that the extinction ratio was only 31\u00C2\u00B14. Problems with theshape of the loss curve which may be due to residual birefringence were also observed. Itis possible that the cell's poor performance is caused by stress from its mounting altering thepolarization of light in it via the elastooptic effect, or that a more precise alignment than ispossible with the current mount is needed. It is also possible that the crystal is simplydefective.In order to improve the performance of the laser, a new mount for the Pockels cellshould be designed, in which the high-voltage connector is attached rigidly to the base of themount rather than to the platform on which the crystal rests, as is done in the current design.Even if the resulting reduction in stress has no effect on the crystal's performance, this47Chapter 4. Conclusions^ 48change would greatly improve the ease and accuracy of alignment, especially if the mountallowed minute rotational adjustments of the crystal about more than one axis.If this change does not bring the laser's performance up to a satisfactory level, thelithium niobate Pockels cell should be replaced by a KDP cell. These cells require a higheroperating voltage and protection from atmospheric moisture, but they have been found in thepast to be very effective electrooptic devices.Another change which should be made at some point is that the spark gap used fortriggering of the laser should be replaced. The new gap should have smaller main-gapelectrodes, with as fine a thread as possible for adjusting the gap spacing. The design of thisgap could also be much simpler than that currently in use, as there is no need to have theconnections for separate circuits for Q-switching and cavity dumping, as in Figure 3.2.Alternatively, an avalanche transistor circuit could possibly be used for triggering in placeof the spark gap. Such circuits have come into common use in laser triggering systems inrecent years, and if transistors with a sufficiently fast switching time are available such acircuit could provide a very practical and robust alternative to spark gap triggering in thisapplication. Compatibility with the spark gaps in the existing CO 2 laser system could beprovided by attenuating the voltage of the trigger pulse they put out and using it to triggerthe avalanche circuit directly. It should be possible to do this without introducing any excessjitter to the system, beyond that of the transistors themselves.Finally, some solution to the problem of longitudinal mode control in an unstableresonator should be found, to improve the shot-to-shot energy stability of the laser. Thismight involve using a tilted plane etalon and then placing a lens at the output to adjust thecollimation of the beam, or some other technique.List of References [1] J. E. Geusic, H. M. Marcos, and L. G. Van Uitert, Appl. Phys. Lett. 4, 182 (1964).[2] W. Koechner, Solid-State Laser Engineering (Springer-Verlag, New York, 1976).[3] A. Yariv, Quantum Electronics, 3rd Edition (Wiley, New York, 1989).[4] P. E. Dyer, in The Physics and Technology of Laser Resonators, ed. by D. R. Halland P. E. Jackson (Hilger\u00E2\u0080\u0094IOP, Bristol, 1989).[5] A. E. Siegman, Proc. IEEE 53, 277 (1965).[6] A. E. Siegman, Lasers (University Science Books, Mill Valley CA, 1986),Chap. 22.[7] F. J. McClung and R. W. Hellwarth, J. Appl. Phys. 33, 828 (1962).[8] W. G. Wagner and B. A. Lengyel, J. Appl. Phys. 34, 2040 (1963).[9] A. A. Vuylsteke, J. Appl. Phys. 34, 1615 (1963).[10] R. C. Weidler, J. H. Burkhalter, and A. A. Vuylsteke, J. Appl. Phys. 38, 4510(1967).[11] H. Houtman and J. Meyer, J. Appl. Phys. 57, 4892 (1985).[12] J. H. Goncz, Instrument Soc. of America, 5, No. 1 [p. ?] (1966).[13] J. E. Murray and W. H. Lowdermilk, J. Appl. Phys. 51, 3548 (1980).49"@en . "Thesis/Dissertation"@en . "1993-05"@en . "10.14288/1.0085109"@en . "eng"@en . "Physics"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "A Q-switched, cavity-dumped YAG laser for use in plasma physics experiments"@en . "Text"@en . "http://hdl.handle.net/2429/1585"@en .