TJCID—21417 Contents DE88 011752

Preface ii Neodymium-Glass Laser Research and Development al LLNL 1

Nova Laser Technology 8 Building Nova: Industry Relations and Technology Transfer IS Managing the Nova Laser Project 26

Optical Coatings by the Sol-Gel Process 34 Frequency Conversion of the Nova Laser 42 Eliminating Platinum Inclusions in Laser Glass ' 52 Detecting Microscopic Inclusions in Optical Glass 64 Auxiliary Target Chamber for Nova 68

DISCLAIMER

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsi­ bility for Ihe accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Refer­ ence herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise dc«s not necessarily constitute or imply its endorsement, recom­ mendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. tAfci Preface

ince the early 1970s, Lawrence construction of the Nova laser Livermore National Laboratory facility. These four articles first S has played a prominent role in appeared in the February 1983 issue the national effort to achieve ot Energy ant! Technology Review controlled nuclear fusion by the (LLNL report UCRL-52000-85-2). inertial-confinement method. The article "Optical Coatings by Working with a series of ever the Sol-Gel Process," from the more powerful lasers, researchers in October 1985 E&TR {LLNL report the LLNL Laser Program have UCRL-520D0-85-10), describes our greatly increased their understanding chemical process for making the of basic laser physics and have damage-resistant, antvreflective silica steadily advanced the limits of laser coatings used on the Nova laser power. Nova, the newest laser glass. The technical challenges in facility at LLNL and the most designing and fabricating the KDP powerful in the world, was dedicated crystal arrays used to convert the in early 1985. light wave frequency of the Nova The articles reprinted here laser are reported in "Frequency record several milestones in laser Conversion of the Nova Laser." Two research at LLNL. "Neodymium- articles. "Eliminating Platinum Glass Laser Research and Inclusions in Laser Glass" and Development at LLNL" recounts the "Detecting Microscopic Inclusions in history of the Laser Program and our Optical Glass," describe how we work on neodymium-glass lasers. dealt with the problem of damaging "Nova Laser Technology" describes metal inclusions in the Nova laser the capabilities of the Nova laser glass. The last article reprinted here, and some of its uses. "Building "Auxiliary Target Chamber for Nova: I-.dustry Relations and Nova," discusses the diversion of Technology Transfer" illustrates the two of Nova's ten beamlines into a Laboratory's commitment to work secondary chamber for the purpose with U.S. industry in technology of increasing our capacity for development. "Managing the Nova experimentation These articles Laser Project" details the appeared in the April-May 1986 organization and close monitoring of E&TR (LLNI. report 52000-86-4/5). costs and schedules during the

Howard Lowdermilk Associate Program Leader Neodymium-Glass Laser Research and Development at LLNL

During the past decade, we have designed and constructed a series of increasingly energetic and powerful neodymium-glass solid- state lasers for research in weapon physics and in inertial confinement fusion (ICF). The twofold focus of our continuing efforts in laser research and development is on (1) the design and fabrication methods that will significantly reduce the cost of a 10-MJ, 500-TW laser system, and (2) the development of techniques for operating solid-state lasers at high average power and efficiency.

For more than a decade, LLNL has to highlight some of the key For further information contact been centrally involved in the nation's accomplishments of the development VV. F. Krupke (4151 J22-53S4. inertial-confinement fusion (ICF) campaign. program. The principal goals of the ICF program are to produce, in the rigin of the ICF Program laboratory, high-gain implosions of the In inertial confinement fusion, fusion fuel, to apply ICF technology and O high-power laser beams rapidly facilities to nuclear-weapon physics heat the surface of a target capsule, research and to military applications, usually containing a deuterium-tritium and, ultimately, to generate cost-effective, fuel mixture, to form a plasma envelope. central-station electric power. In direct The rocket-like blowoff of plasma support of this program, we have material from the surface drives the developed the technology base and capsule inward to compress and heat the increased the energy and peak power fuel. When the core reaches a density of levels of neodymium-glass lasers by 10' to 104 times that of liquid water and several orders of magnitude. On the a temperature of 10K degrees (10 keV), occasion of the initial operation of the the fuel ignites. Thermonuclear bum most recent of these laser systems. Nova, spreads rapidly throughout the it is appropridle to review the strategic compressed and inertially confined fuel, technical issues of inertial fusion that we yielding many times the driver input addressed in the early 1970s, to outline energy. Shortly after the invention of the the considerations that led to the ruby laser in 1960, computer calculations selection and subsequent development were made at LLNL to simulate the of neodymium-glass systems for ICF irradiation of tiny deuterium-tritium and weapon-physics research, and (D-T) pellets by .ritense pulses of laser light and their subsequent implosion to U.S. inertial fusion effort. At that time, thermonuclear conditions. A! this same both laboratories were engaged in time, it was recognized th.ll the fusion exploratory development of several types microexplosions could eventually he of lasers for possible use in inertial applied to the generation of power. confinement fusion, including carbon Calculations revealed, however, that dioxide, neodymium-glass. hydrogen efficient generation of fusion energy fluoride, and atomic iodine lasers. would not result from simple laser Because it was judged that all of these heating of the thermonuclear fuel. lasers could, with sufficient development, Instead, to generate fusion energy, lasers be scaled to the required energy and would have to compress and implode peak power levels, selection of which the fuel to 10000 times its liquid density. types to pursue was centered on the Experimental work to create hot operating wavelengths of the lasers and plasmas using lasers began in the mid- on their potential for high average power 1960s al several laboratories including and efficiency. LLNL Neutron generation from planar Analyses of plasma physics, coupled lithium-deuteride targets was reported with LLNL computer calculations, by Soviet researchers in 1968. With this indicated that i( would be highly demonstration of neutron generation beneficial to deliver the bulk of the from laser-heated plasmas, interest energy to the target at shorter intensified in seeking an answer to the wavelengths'; at a short enough basic question Can we design fusion wavelength, the laser radiation incident pellets that will produce useful on the fusion pellet would be absorbed thermonuclear gain {that is, release primarily through inverse during thermonuclear fusion more than bremsstrahlung, and a thermal plasma 100 times the energy used to implode would be created. These computations the fusion fuel) when driven by also enabled us to identify and technologically and economically characterize a number of undesirable practical lasers? competing energy-absorption processes The best available computer-based that, at longer wavelengths, would lead estimates for laser-driver requirements to nonthermal plasmas and to deleterious with gains of 100 were many orders of preheating of the fusion fuel. Analyses magnitude beyond the energy and peak of the activation thresholds for these power levels of lasers available in the detrimental absorption mechanisms earlv 1970s. Thus, it became evident that indicated that their thresholds would be the minimum drive requirements could higher when shorter-wavelength laser (and should) be determined using light was used. experimentally flexible, single-shot laser Although we could not. at that time, systems, and that development of the project quantitatively the laser intensity efficient, repetitively pulsed laser systems at which these unwanted processes needed for the more demanding fusion would occur, it was clear that by applications could (and should) be selecting a laser driver with the shortest undertaken separately and at a rate operating wavelength, the physics risk in consistent with growing knowledge of driving efficient dense implosions would the minimum drive requirements. be significantly reduced. Thus, on the basis of wavelength considerations alone, the neodymium-glass laser became the election of Lasers for ICF laser of choice because it has the shortest In 1971 and 1972, tlv U.S. Atomic operating wavelength of 1.05pm SEnergy Commission and its (compared to 10.6 j/m for carbon dioxide, nuclear weapons laboratories at 2.70 /urn for hydrogen fluoride, and Livermore and Los Alamos 1.315,um for atomic iodine). Anticipating recommended to Congress that dedicated that even a 1,05-um wavelength might irradiation facilities capable of producing not be short enough, we noted that more than 10 TW of peak power should small-scale laboratory experiments had be undertaken as the next step in the demonstrated the scientific feasibility of INERTIAL FUSION

using nonlinear harmonic-converter of special nuclear materials for military crystals to efficiently shift the 1.05-jtm applications, breeding fuel for light-water output radiation of a neodymium-glass reactors, and producing electric power for laser to green (0.525-jtm), blue civilian applications). (0.350-pm), and ultraviolet (0.265-/vm) light. eveloping the In addition to these desirable Neodymium-Glass Laser characteristics, the neodymium-glass D In 1972, the principal elements laser offers the flexibility in pulse width of the neodymium-glass technology base and pulse shape required for a research- were limited to (1) a series of oriented irradiation system designed to neodymium-glass rod amplifiers and explore a wide variety of target designs systems designed by Compagnie and physics phenomena. The efficiency Generate d'Electricite in France, (2} face- and average power potential of the pumped disk amplifiers of approximately neodymium-glass laser, however, was 10-cm aperture developed at LLNL and judged to be inadequate for the projected at the Naval Research Laboratory, and laser-system requirements for civilian (3) a relatively high-performance lithium applications (e.g., power generation). Of silicate laser glass (ED-2) developed and . the types of laser considered, the carbon manufactured bv Owens-Illinois of dioxide laser looked to be the other most Toledo, Ohio. Laser systems of the day promising laser, offering the potential for were able to generate peak power high efficiency and average power but outputs of approximately 300 GW in a operating at a relatively long wavelength. 1-ns pulse. Because, in the early 1970s, it was From this starting point, we launched uncertain just how short the irradiation a broadly based, long-term effort into wavelength had to be to achieve a high- the research and development of gain implosion at a pijviical laser energy neodymium-glass lasers. In the 13 years and power level, and because the since then, we have designed and development of ICF technology would constructed a sequence of neodvmium- necessarily take several decades, a glass laser systems at LLNL, culminating decision was made to explore more than in our largest and most powerful system one research and development strategy. ever, the 100-TW Nova laser. Figure 1 Within the U.S. ICF program, the Los summarizes the energy, peak-power, and Alamos National Laboratory elected to pulse-width capabilities of the Janus, pursue the long-wavelength approach to Cyclops, Argus, Shiva, and Nova laser ICF, with the objective of exploiting the systems. The Nova laser will be capable efficiency and average power potential of the carbon dioxide laser. Los Alamos has since successfully developed carbon dioxide laser technology to the 40-kJ, 50-TW level with their Antares laser system brought on line in 1984. We at LLNL decided to pursue the short-wavelength approach to ICF with a two-pronged laser research and development strategy. Neodymium-glass laser systems (fitted with harmonic converters, if necessary) would be developed in a sequence of experimentally flexible, single-shot, Fig. 1 fusion-target irradiation facl'ties. Energy, peak-power, and pulse-width Concurrently, we would search (or and capabilities of the neodymium-glass develop new, short-wavelength lasers laser systems constructed and oper­ with the efficiency, average power, and ated at LLNL during the pasl decade. The Nova laser system can operate at cost potential required for high-average- 10 1 1 10 105, 0.525, and 0.35 pm; the others power JCF applications (e.g., production Peak power, TW operate only al 1.05/an.

3 of generating output radiation at 0.525 beam, which arises from the intensity- and 0.350 /im, as well as at its dependence ol the nonlinear refractive

fundamental 1.05-pm wavelength. index (n;) of optical materials. The The Shiva laser (10 to 25 TVV) was second phenomenon is the catastrophic a S25-million line item authorized by damage that occurs at high beam Congress in 1973 and operated in 1977. fluences to the optical components It was the technological fulcrum of our locatpd in the beam path. We initiated entire laser-development enterprise. The research efforts to find laser glasses and results of the research and engineering other optical materials with lower

efforts undertaken to create the Shiva nonlinear refractive indices (n;), to laser were so successful that it became develop beam control and propagation possible to construct neodymium-glass techniques to minimize beam laser systems with peak power and degradation in the presence of a

energy levels orders of magnitude greater nonvanishing n2, and to develop optical than Shiva's. These advances also made materials with significantly higher possible significant reductions in the unit damage threshold fluences. cost of laser output energy in each succeeding, larger laser system. The New Optical Materials technical issues and research results Following the development of a leading to the realization of such theoretical model relating n to the Fig. 2 : dramatically improved system material's linear refractive index (n) and Location of various silicate, phosphate, performance levels have been discussed to its dispersion (expressed by the Abbe fluorophosphate, and fluoroberyllate in the technical literature.2 laser glasses in the refractive number, i;,), rapid progress was made in index/Abbe-number plane. (Abbe num­ developing optical materials with lower ber is the reciprocal of the material's erformance Limitations n: values. Experimental rr, data on a index of dispersion.) The locations of There are two phenomena that wide variety of glasses confirmed the several fluoride crystals also are quantitative predictive value of this shown. Curves of constant nonlinear re­ still limit the performance of fractive index (nj) are indicated by the neodvmium-glass lasers. The first is the model. As shown in Fig. 2, curves of dashed tines. small-scale self-focusing of the laser constant n2 can be overlaid on the linear- index/dispersion plane. Optical glass companies routinely report the linear index and dispersion of the many 1.9 thousands of glass compositions they / formulate, enabling us to obtain readily

the n: values for these glasses. Thus, we 1.8 \ / were able to target for development a \ / number of glass compositions \ / characterized by n values lower than " s / 2 Silicates, \ / those of the silicate glasses (e.g., ~ phosphates-v 1( *v \S \ »FH-5 phosphates, fluorophosphates, and fluoroberyllates). h.c ^. .V\5,. We also developed a theory of ~ Fluorophosphates -^ ^^ ^ /lU««*_ LHG-91H •'M\ neodymium-ion transition probabilities to \ X / ^^3r / relate the laser properties of a glass to - r FJuoroberyl fates J>cv ( sr \/ E1.5 its composition. This model, together FK S V , \ "*** -^ ^ «FK-51 • *\/ •>l^>^^ ^^ with the linear-index/dispersion model, CaF \ 2 J^K ^» LHG-10 ^Sr^ v provided us with a powerful predictive 1.4 \ .i^f^L^L^L^La tool for identifying new neodymium- ^^^^^f*^^\. ^- Fluorosilicates \ . _ KJ glass compositions with substantially f»^^-E-1A1, ""^ superior laser performance capabilities. Glass manufacturing companies (Owens- 1.3 -I LG-812 *"• •*, ~^^^^„. "**(% O.S Illinois, Hoya Corporation, Schott - BeF, (BB00) """-/% 0.2S Optical Company, Kigre, and Corning) participated throughout this effort, 1.2 , 1,1,1,1,1,11 110 100 90 80 70 60 40 providing expertise on glass-forming Abbe number, IU science and production.

4 INERTIAL FUSION

Beam Control illustrated in Fig 3, which shows the A major advance in laser beam fivefold increase in the damage threshold control was achieved in 1976 with the fluence of antireflecting surfaces achieved introduction of the relay-imaging spatial in collaborative research efforts with filter. Essentially a telescope with a Corning. Owens-Illinois, Schott Optical diamond pinhole at the common focus Company, Hoyn Corporation, and of the lens pair, this device performs two Westinghouse. Because antireflecling separate functions when it is placed in a surfaces typically are the most heavily chain of laser amplifiers. First, it expands stressed optical surfaces in laser systems, the diameter of the laser beam as it the development of these high-damage- passes from smaller diameter amplifiers threshold optical components has had a to larger ones, maintaining the beam profound impact on high-performance fluence along the chain below the lasers. fluence thresholds for optical damage. Second, it removes small-scale intensity horter Wavelengths and phase noises in the laser beam as for Nova the beam passes through the pinhole, SThe sequence of LLNL laser preventing the build-up of deleterious systems (see Fig. 1) has been used, over hot spots in the beam. With this spatial- the years, to conduct a large number of filtering technique, the peak-power laser-plasma interaction and implosion output of the Argus laser system was experiments. Experiments with the Argus increased from 2 TW to more than 4 TW and Shiva lasers showed that ].06-pm using simple relay-imaging optics. The light produced a normegligible number principle of relay-imaging spatial filtering of the hot electrons that cause is now incorporated, worldwide, in the deleterious preheating of the fusion Fig. 3 design of all high-power neodymium- fuel. These and similar observations A fivefold improvement in the damage- glass laser systems. made at other fusion-research th.-eshohi fluence of antireflecting sur­ laboratories prompted fusion researchers faces has been achieved at LLNL in the to undertake similar measurements at past few years (incident laser wave­ Optical Damage length of 1.06 im, pulse length of 1 ns). The key to our success in reducing shorter wavelengths, converting the Because antireflecting surfaces are the damage to optical materials was the 1.05-fJm fundamental wavelength of most severely stressed optical sur­ decision to build high-quality laser- existing neodymium-glass lasers to faces in high-performance laser sys­ shorter wavelengths. Experiments using tems, development of such coatings irradiation facilities and beam diagnostics with high damage thresholds has made with which to make accurate and green, blue, and ultraviolet light showed a strong decrease in the hot-electron a significant impact on the performance repeatable damage-threshold capabilities of laser systems. measurements. We worked closely with optical fabrication companies (Eastman Kodak, Zygo Corporation, Tinsley Laboratories, and Perkin-Elmer Corporation) and optical coating Laser-cleaned, neulnjt-sokitkxi, gradient-index surface companies (Optical Coating Laboratory, 25 Inc., Spectra-Physics) to prepare large numbers of substrate and coated elements (antireflection, high-reflection, polarizing) under varied but reproducible 20 conditions. Damage-threshold measurements made at LLNL were used to identify the most promising optical- 15 element designs, materials, and - Thin-film coating As-prepared, neutral-solution, _ on conventionally production methods. Because this S'lO gradient-index surface research and development process was E polished substrate conducted directly with the optica! • component manufacturers, the results Thin-film coating could be translated into production optics on supetpolished substrate with minimal delay and cost. A measure of the success of these efforts is D 1977 1978 1979 1980 1981 1982 1983

5 fraction as the wavelength was because high average power could be shortened, particularly between 1.05 and realized by convectively removing the 0.525 fim. The fraction of incident laser waste heat. In addition, in the early light absorbed and the ablation pressure 1970s, only very poor efficiency and produced aiso increased significantly at average power performance had been shorter wavelengths. On the basis of demonstrated for neodymium-glass and these highly favorable results, we revised other solid-state lasers. the Nova performance specifications Between 1974 and 1979, a variety of from the original 250kj of ].05-/im light energy-storage and steady-state gas to the more target-effective specification lasers were proposed, developed, and of 100 kj of 1.05-^m light with assessed for use in fusion applications. conversion to green and blue light. The KrF excimer laser, fitted with one of several forms of pulse compression, eyond Nova proved to be the best of the systems Our research and development investigated. After further study, B efforts have been so successful however, the KrF laser in its present that we are now assured that short- form was judged to be marginal in wavelength laser systems can, indeed, efficiency (below 6%), extremely complex be scaled to 10 MJ and 500 TW. These optically because of the need for pulse- energy and power levels should be width compression, and several times too sufficient to drive the high-gain fusion expensive for the end-use applications targets required for the more demanding of ICF. We are continuing to seek a single-shot weapon-physics research and breakout from these limitations on KrF military applications. and other excimer laser systems. If, however, such a laser system were At the same time we are reexamining to be constructed with Nova technology the potential of solid-state lasers to and the current master-oscillator/power- operate efficiently in a repetitively pulsed amplifier (MOPA) system architecture, mode. Our analyses3 are based on new the cost would be prohibitive. solid-state laser design concepts and new Consequently, for the past year or so, we pumping sources, host materials, and have been engaged in an effort to design laser ions. With these advances, and assess new cost-effective system unconstrained scaling of average power architectures, to develop new higher- while maintaining high beam quality performance laser materials, and to does appear possible if we use an develop large-capacity, low-cost amplifier geometry consisting of thin components. Although it is still too early slabs of gain medium that are flow- to be certain that 10-MJ-class lasers can cooled on their two large surfaces. be constructed for the budgetary goal of Furthermore, with careful conservation $25 to $50 per joule, large reductions in of pump photons by a variety of well- some of the major cost items appear identified techniques, efficiencies in possible, and we are continuing our excess of 10% should be possible at efforts to obtain similar reductions for high power and radiance. We are now those items that are still driving up the engaged in an effort to validate these system cost. ideas and to thoroughly assess the new materials and pump sources. ther Short-Wavelength Lasers CF Program Philosophy OThe second prong of our laser The fast-paced (10-year) research and development strategy has I development and deployment of focused on those short-wavelength lasers highly functional neodymium-glass that have an efficiency, average power, target irradiation systems exceeding and cost suitable for the proposed ICF 100 TW is a singular accomplishment by applications requiring repetitively pulsed any measure. Several key institutional, operation (e.g., power generation, fuel programmatic, and managerial factors production). These research efforts have allowed this to occur. First, our initially were centered on gas laser media laser research and development has been conducted in the context of a "full Defense). It was thus possible to spectrum" program. That is, the users capitalize rapidly on research advances of the laser technology and systems while the laser-system facilities were being developed are actively involved— being developed and constructed. organizationally, programmatically, Lastly, especially at LLNL, heavy use geographically, ar ' psychologically—in has been made of computer modeling, the development process itself. This has off-line facilities for development and minimized feelings of underachievement testing, and extensive diagnostics. by the laser developer and Significant commitments to the overexpectation by the user. It has development of new materials also were also ensured that both parties share made. We have realized a manyfold a common commitment to the success return on these investments in terms of of the program as a ivhole. laser-system performance, schedule, and Second, long-term laser development cost. These policies, prerogatives, goals were clearly established at the and practices operating within the outset. Very early on, we examined the DOE-LLNL system represent an physical phenomena involved in inertial enormous strength as we continue to fusion, identified those of highest pursue the goals of the national ICF leverage, and identified, to the best program. of our knowledge, their parametric dependences on laser-radiation onclusion parameters. Requirements for end-use As a result of our 13 years laser systems also were estimated (e.g., C of intensive research and reactor drivers). Because of our limited development, our innovative approach to knowledge, these analyses were quite program management, and our effective judgmental. They did, however, provide working relations with industry, we have us with a very valuable framework established the premier JCF fusion against which to compare the potentials research facility (Nova) in the world of the laser media then known, to select today. Additionally, we have denned a primary and back-up approaches, to technological pathway extending from engage in a comprehensive research and single-shot, neodymtum-glass, fusion- development effort, and to establish research lasers to efficient, high-average- programmatically rational intermediate power, solid-state drivers for future ICF and ultimate goals. fuel- and power-producing Third, our most realistic estimates applications. L3 of technical risk, cost, and schedule to accomplish various laser milestones were presented directly to our sponsors Key Words: inertial confinement fusion (ICF)— (initially the Atomic Energy Commission, research ,ind development: laser—Argus, Cyclops. Krl-'. (anus, neodymium-glass, Nova. 5hiva. then the Energy Research and short-wavelength, solid-state. Development Administration, and now the Department of Energy). Commitments were made to specific Notes and References milestones on the basis of available I. ]. Nuckolls, I.. Wood, A Thiessen, and and adequate funding. Resources were G.Zimmerman ,\jlure. 239. 139-M2 (1«J72) expended as needed throughout the 2 f. r. Hol/nrhter. vt ,?/.. Laurence Livermore research and development efforts \almnal laboratory, Rept. UCRI- 52868 without placing artificial constraints on the actual dollar amounts allocated to 3 II. Emmelt. IV f1 Krupke. and IV. R Sooy Trie Potfntijt of Hfe,ri-.-Uerii;e-Aitver Solid research versus to development (as is ( 5t,ife /iivrv l^wrence Livermore \'atinna] often done in the Department of Lirnraiciry. Repl. LCRI.-53571 <1W| Nova Laser Technology

In building the Nova laser, we have significantly advanced many technologies, including the generation and propagation of laser beams. We have also developed innovations in the fields of alignment, diagnostics, computer control, and image processing.

For further information contact Nova is the world's most powerful at least 10 to 30 times more energetic John F. Holzrichter (415) 423-7454. laser system. It is designed to heat and than Shiva would be needed to compress small targets, typically 0.1 cm investigate ignition conditions and in size, to conditions otherwise produced possibly to reach gains near unity. only in nuclear weapons or in the Because of the importance of such a interior of stars. Its beams can fa.ility to the progress of inertial fusion concentrate 80 to 120 k| of energy (in research and because of the construction 3 ns) or 80 to 120 TW of power (in time entailed (at least five to seven 100 ps) on such targets. To couple ene-gy years), the Nova project was proposed to more favorably with the target. Nova's the Energy Research and Development laser light will be harmonically converted Administration and to Congress. It was with greater than 50% efficiency from its decided to base this system on the near-infrared fundamental wavelength proven master-oscillator, linear-amplifier- {l.Ob-fim wavelength) to green (0.525- chain laser system used on the Argus pm) or blue (0.35-Jlm) wavelengths. The and Shiva systems. We had great goals cf our experiments with Nova are confidence in extending this to make accurate measurements of high- neodymium-glass laser technology to the temperature and high-pressure states of 200- to 300-kJ level. As target-physics matter for the weapons program, to data were obtained with the Shiva compress deuter.um-tritium (D-T) fusion system and as further computational fuel to densities .ipproaching 200 g/cm3 modeling occurred, it became clear that (1000 times the density of liquid D-T), to the most important demonstrations for determine the ultimate energy required the Nova system would be to generate for efficient ignition and burn of inertia! a very high-quality compression fusion targets,1 and finally, to make environment with short-wavelength important physics and engineering (<0.5-/jni) laser light and to determine measurements such as x-ray laser more exactly the laser energy required studies. Figure 1 shows the Nova laser for high-gain fusion reaction.1 facility. As a result, the final Nova laser The Nova concept was born while configuration is a ten-beam system able LLNL's 10-kJ Shiva laser was under to provide light output at 1.05, 0,53, and construction in 1975. Physics experiments 0.35 pm over a wide pulse-duration then being conducted on the range (0.1 ns [o greater than 100 ns). Laboratory's Arg-js laser, together with It incorporates a flexible target theoretical analyses, indicated that a laser configuration for inertial-confinement INERTIAL FUSION

fusion (!CF) direct drive. 1CF indirect • Maximum flexibility to permit the diive, x-ray laser, and other target widest variety of weapon physics, ICF, applications employing two target and general physics applications. experimentation areas. These goals were successfully met. As discussed in the article on p. 1, we Two beams of the Nova laser, opprating had gained a great deal of laser physics as the temporary Novette facility, have and engineering experience ivith earlier verified most of the design of Nova's Fig. 1 laser systems—Cyclops, Argus, and laser chain and have demonstrated many Cutaway view of the Nova laser fusion Shiva. The philosophy that guided our of the expected benefits of short- facility. The space frame at the right Nova design efforts reflected these wavelength target irradiation. supports the ten laser am/. Met chains. experiences. The resulting design A system of high-refleclvity mirrors incorporated: ova Laser Design causes the ten laser beams to arrive 1 simultaneously at the fusion target, • As large an output aperture as The Nova laser system" is which is centered in the spherical tar- possible to reduce beam count and N shown schematically in Fig. 2. gel chamber near the top left of the thereby system construction and It consists of a master oscillator and ten picture. maintenance cost. • Maximum fluence through each aperture. This required the development of new laser glasses, propagation techniques, and coatings to sustain the high laser intensity. • Automation of beam alignment and electronic system testing to reduce operating costs and increase the shot rate. • Flexibility to accommodate new technology that would increase performance. • Reuse of Shiva components to reduce cost wherever possible.

r^y~ Output sensor r^X^n Spatial filters < package ^H Amplifiers

Isolators

| | Space for added amplifiers

\ / Turning mirrors 46.0-cm diameter

Oscillator

Splitter

Path equalization

31.5-cm diameter

Arrangement of the major optical com­ ponents in a representative Nova beam line. Space provided in the two larger amplifier sections makes il possible to add more amplifiers and increase the beam power at low cost

9 chains of laser amplifiers {only one of laser research to 100 ns for weapon- which is shown). The master oscillator effects simulations) and shape (for generates a low-level pulse, typically example, a series of 1-ns, "square" 0.1 m], which is preamplified to about pulses, spaced by 5 ns, each more 10 J. The pulse is split into ten beams energetic than the previous one by a and repeatedly amplified and expanded, factor of 10 to 50). The pulses are until, at the exit of the laser chain, amplified in a series of glass-rod each beam is about 10 kj and 74 cm in amplifiers of increasing diameter (each diameter. Each beam is then directed with a gain of 20 to 25) until the beam to the target chamber by four or five reaches 4 cm in diameter. Spatial-filter turning mirrors. Before the beams enter telescopes further expand the beam to the target chamber, they are each 10 cm, the diameter of the first disk- diagnosed by sampling a fraction, amplifier section. Each disk-amplifier harmonically converted by passage section has a small signal gain of 4 to 10. through nonlinear KDP crystal arrays, The advantage of the disk amplifiers is and finally focused to a spot on the that they can be scaled to almost target 0.25 mm in diameter. Figure 3 is a arbitrary beam diameter and still Fig. 3 view down Nova's ten amplifier chains. maintain the ratio of high gain to low Nova amplifier chains mounted on their The master oscillator system1 glass thickness that is required for good space frame. Each frame hotels five generates a time bandwidth-limited laser beam propagation. As the beam passes laser chains, each of which is 137 m from one amplifier stage to the next, long. pulse of variable width (0.1 ns for x-ray it is expanded to 15 cm, 20 cm, 31.5 cm, 46 cm, and finally 74 cm for propagation ^35^ to the focusing lens. After each stage (where the gain is a factor of about 10), an isolator or one­ way transmission element is required.'' Otherwise, parasitic reflections from shiny metal parts in the laser chain or from the target cause the laser to break into spontaneous oscillations and cause damage to sensitive components. For the small-diameter sections (up to 5 cm), we use a Pockels-cell isolator. The isolator, which acts as a "light switch," consists of an electro-optical crystal between crossed polar.zers. By allowing light to pass for only 10 ns, it prevents the build-up of parasitic oscillations and keeps unwanted, spontaneous emissions generated in the front end of the laser from being amplified to a large enough level to damage the target. All the large- diameter stages contain Faraday-effect isolators. These devices rotate the polarization plane of backward-reflected light in a direction opposite to that of the forward-propagating laser beam. A polarizer plate in the path of the backward-reflected light rejects the parasitic modes. Most of the isolator components were developed and tested on the Shiva laser system and are being reused on Nova. Spatial-filter telescopes have three primary purposes: (1) to expand the

10 INERTIAL FUSION

diameter of the beam, (2) to smooth it eam Propagation with the filter, and (3) to project it The performance per unit cost down the chain to prevent unwanted B of any laser system increases diffraction. A spatial filter consists of rapidly with increasing laser fluence." a pair of lenses and a pinhole located Most of the costs of these lasers are in a vacuum tube between them at the either somewhat independent of aperture lens focal points. The first lens focuses (fixed) or proportional to beam area the parallel-propagating rays of the (aperture size). Fixed costs are associated laser beam through the pinhole. Any with the building, the target systems, misdirected rays (the cause and result the space frame, the computer, and the Fig. 4 of local beam nonuniformity) come to a diagnostic system. Variable costs that are focus either nearer to or farther from the proportional to beam area are associated Interchain pea* and average fluences fiTSt lens and therefore do not pass as a function of distance along the with optical elements, amplifiers, spatial laser chain for e 10.5-kJ laser pulse through the pinhole located at its filter, isolators, etc. By increasing 1 ns long. The sizes and locations of focal point. The second, larger lens the beam fluence through the baseline the spatial filter lenses are indicated recollimates the beam diverging from the system, we obtain more total energy along the top of the figure. Entrance- pinhole to match the larger diameter of with essentially the same components. lens surfaces will suffer damage if the the next amplifier section. This optical Major limits on beam power (discussed peak fluence exceeds the indicated limits. system also arts as a telescope to project and magnify the image of the beam as it 20 passes to a larger-diameter section KDP array - without creating deleterious diffraction ripples. — *-*+«—f>^-

Sol-gel antireflecrjng This laser architecture, consisting 15 Most threatened optic of a stries of amplifiers of increasing coating damage diameter, prevents damage to any of the threshold many optical surfaces by light of too high an intensity. At the output ends of 10 the larger amplifier sections, the peak energy fluence approaches the damage threshold for uncoated glass. The most threatened components in the laser are s _ Polarizer damage the input lenses to the final spatial filters. threshold After the beam has been expanded and smoothed by the spatial filter, its peak energy fluence is well below the damage threshold for coated lenses and safely below the damage threshold of the Distance along laser chain, m harmonic conversion system at the output (Fig. 4}. The performance of this propagation strategy was modeled in detail by the fast Fourier-transform code MALAPROP.' The predictive success of this comprehensive code is the result of the fine mesh size permitted by LLNL's Cray computers, the accurate modeling of diffraction and saturation, and our spatial noise model, which mimics the Fig. 5 effects of small scattering sites and Full computer simulation (using the correlates well with independent MALAPROP code) displaying the uni­ statistical observations of noise sources. formity of a 12.8-kJ laser pulse 3 ns Figure 5 is a full computer simulation of long as it appears entering the the variation of the uniformity of a harmonic conversion array at the target chamber. Profiles toward the back of 2.50 12.8-kJ beam with time over the life -40-20 0 20 40 the figure are the first to arrive at the of a 3-ns pulse. Beam diameter, cm target.

11 in the article beginning on p. I) are A mplifiers catastrophic damage thresholds and L^L Disk amplifiers and other cumulative nonlinear, self-focusing J. A. components through the aberrations. Important engineering 15-cm-diam stage are similar to (or concerns related to aperture usage reused) Shiva laser components include the efficiency with which the Although optimized for Shiva's shorter aperture is filled (the beam fill factor), pulse operation, these components are the transmissivity of optical elements, acceptable in these smaller aperture and the efficiency of the harmonic sections of Nova. The output amplifier conversion. Additional issues affecting stages (the 20-, 31.5-, and 46-cm-diam cost are the beam amplification efficiency amplifiers) were specially designed for (energy extracted from the gain medium) the K'ova laser to improve electrical and the electrical efficiency with which efficiency and to meet propagation amplifier gain is produced (discussed constraints on the ratio of gain to giass below). thickness. A goal of the research program Figure 6 shows one of the 46-cm-diam supporting the Nova laser was to amplifiers in its partly assembled maximize each of these influential rectangular housing. The two elliptical propagation parameters. Technologies glass disks are tilted at Brewster's angle developed to this end include new to permit almost zero transmission loss. phosphate laser glasses characterized by The disks are paired to compensate for Fig. 6 high gain and a low nonlinear index of beam offset caused by refraction in each refraction, new neutral-solution and tilted plate The amplifier housing is Partly assembled rectangular laser- disk amplifier (46-cm diam aperture). sol-gel antireflection coatings and lined on the top, bottom, and both sides The amplifier disks are elliptical for over-coated thin-film reflectors, new with silver-plated reflectors to provide placement at an angle to the circular apodizarion techniques to operate with a tight optical coupling between the disks beam. Each disk is divided in halt (indi­ fill factor of over 80%, new antireflection and the flashlamp-- One set of the cated by the vertical band) and clad coatings to reduce transmission loss in flashlamps, backec. by a silver reflector, around its edges with colored glass to the crystals used in harmonic conversion, is visible in Fig. 6 L ..'hind the two minimize unwanted spontaneous laser action, which would reduce the amount efficient (up to 80%) harmonic- elliptical glass disks. A shield of flat of energy available lor amplifying light conversion techniques, and new beam- glasi between each row of flashlamps traveling in the beam direction. Trans­ focusing techniques to minimize the and the amplifier disks helps protect the verse flashlamps and their reliectors number of lens surfaces and their disk surfaces from damage should a appear at the rear. The interior parts ol complexity for multiwavelength target flashlamp explode. These shields also the optical cavity are silver-plated for high reflectivity. focusing. preserve beam quality by keeping turbulent convection currents of flashlamp-heated air out of the beam path. Scrupulous clean-room techniques are used throughout the assembly, maintenance, and operation of these amplifiers (the technician's costume is lint-free) to keep the optical surfaces free of scattering centers, such as minute dusk specks, which are the main source of beam nununiformity and of potential optical damage. The aperture in the final 46-cm amplifier is so large that it posed two special problems: the possibility of spontaneous internal laser action within the plane of the disk and a size exceeding that for economical fabrication. To solve both problems, we split the disk in half (the vertical band visible across the middle of each disk in Fig. 6) to shorten the transverse gain path and to

12 INERTIAL FUSION .

reduce the size of each plate. We then fabrication restrictions while maintaining clad the edge of each half disk, all the a desirable ratio of gain to glass wav around, with a copper-doped glass thickness. that matches the index of refraction of the disk's phosphate glass. This edge omputer Controls cladding absorbs spontaneously emitted Without an extensive and laser light before it can undergo C sophisticated computer control reflection and be amplified to a network, it would be impossible to significant degree. Left uncontrolled, control a complex la:-er system such as such a buildup of spontaneous emission Nova, with its 1600 high-power electrical would deplete the excited-state circuits driving 5000 flashlamps and population in the glass before the desired 32 Faraday rotators, together with pulse arrived to sweep the energy from hundreds of moveable mirrors, electronic the amplifier. control functions, and diagnostic sensors. A split disk produces a split beam, as Figure 7 outlines the Nova control seen in the beam profiles in Fig. 5. By system, in which fifty LSI-11/23 Fig. 7 using apodizing spatial filters and image- microcomputers are linked with four Diagram of the Nova control system, relaying techniques, we can control VAX-11/780 computer systems. Common showing the use ol multiple and hardware and software provide the interchangeable minicomputers potential diffraction problems associated (VAX-11/780) and microcomputers with obscuration of the beam's center flexibility to transfer operations and (LSI-11/23). The operator control con­ line. Segmented-disk technology is functions from one computer to another soles, also interchangeable, are ol the crucial to Nova and to future laser in the network. touch-panel type. We use optical fibers systems because it allows us to expand, extensively to avoid electrical interfer­ Nova's control system was designed ence tram high-current discharges (as almost arbitrarily, the diameter of a to satisfy control and data-acquisition trom flashlamps) when transmitting glass disk amplifier without material- functions in four areas: computer data and conlrol information.

Operator Hardcopy consoles

VHX analysis computer To VAX BuMing381 development system VAX confol computers Video One-way data analysis system

HLow-spee d

Muftiport memory Optical liber

LSI-11/23 Event libers ••Pink" "Red" logger Power- — -A >— video video Novalink conditioning • sources sources Device/ fibers devices cameras —< ) *** *K>

r_ €3 Alignment and diagnostic devices and computers

13 • Power conditioning, including images, calorimeters, high-voltage capacitor-bank activation, laser firing, power supplies, interlocks. 20-kV and system timing. digitizers, transient digitizers, and • Alignment of the individual laser remote image memories). The control components and the target. required for these elements ranges from • Laser diagnostics (measurements simple status monitoring to the of beam energy and uniformity) demands of closed-loop alignment • Target diagnostics (measurements through the digital image processing of of temperaUi-.j and densities achieved two-dimensionai beam profiles. in the target implosion). As an example of the system's remote, When designing the Nova system closed-loop alignment capability, in 1979, ive made use of many new consider the task of positioning each of advances in digital controls. These over 70 spatial-filter pinholes scattered included high-speed (10-MHz) optical- throughout the laser bay. If done fiber communication links that manually, this task ivould require a crew interconnect the distributed computers of technicians and would take several and devices, an advanced control-system hours. In our system, a solid-state language, named Praxis (developed vidicon camera forms an image of the jointly with Bolt. Beranek. and Newman, output beam and sends it to the control Int.). high-speed parallel processors for system. The control system automatically automatic image processing and beam processes the image and issues positioning, and touch-sensitive panels commands to stepper motors throughout overlaying color graphics displays on the the laser bay to adjust the positions of operations control consoles. the various pinholes, completing the The control system was organized alignment in only 30 minutes. into four main contro! and data acquisition subsystems and a fifth unifying subsystem (central controls) armonic Conversion that integrates functions and centralizes A crystal of potassium operations. The system supports more H dihydrogsn phosphate (KDP), than 5000 individual control elements properly cut and oriented in a laser and sensors (stepping motors, video beam, can convert up to 80% of the incoming light to a second or higher harmonic, thereby increasing its frequency." However, applying this techniquf and diagnosing the results on beams of the size and intensity of Nova's required considerable innovation, Conventional approaches to the design of harmonic converters capable of g?nerating both second- and third- u*ir han.ionic light would require three assemblies containing crystals of three different thicknesses. We developed a system that consists of a single crystal array1 containing nine sets of two crystals positioned back to back (Fig. 8). Fig. 8 This arrangement increases the interchangeability of parts and An assembled harmonic conversion ar­ ray of KQP crystals sandwiched be­ significantly reduces costs. To generate tween transparent windows of fused the second-harmonic frequency, the silica. Two of these arrays, appropri­ array is oriented so that the two crystals ately aligned and irradiated with polar­ function independently, producing the ized 1.05-/im light from lhe laser, on harmonically converted light in two provide either 0.53-^m (green) or 0.3S-;irn (blue) light for more effective orthogonally polarized components, one coupling to the ICF target. from each crystal. To generate the third

£ 14 INERT!AL FUSION

harmonic, we simply rotate the assembly neutron and x-ray target yields, and it about the beam direction by O.I7rad includes a basic set of target diagnostic (10 deg) and turn the crystals about instrumentation. 6 mrad (0.25 deg) to the proper phase- Nova's primary target chamber is an matching angle, thereby converting two- aluminum sphere 4.6 m in diameter. Its thirds of the incoming infrared (1.05-//m) diameter was originally chosen to beam to green (0.53-/im) light in the first support 20 final-focusing f/4 lenses 74 cm crystal. The second crystal mixes the in diameter at their correct focal distance remaining infrared light with green light (Nova was originally designed with 20 to produce third-harmonic blue light laser beams). However, as the final (0.35 ^m). Efficient conversion is ten-beam design with harmonic- obtained over a wide range of input conversion arrays requires the same focal beam intensities for second-harmonic distance, it was not necessary to change generation and over a smaller but still the chamber size. The walls of the adequate range for third-harmonic chamber are 13 cm thick to support generation. Figure 9 shows a atmospheric pressure, the weight of the harmonically converted Nova-like beam crystal arrays, and the focusing-lens drive produced by the Novette laser. mechanisms while maintaining accurate The size of Nova's beams poses alignment. The chamber is made of special problems for harmonic aluminum to prevent the buildup of conversion. Growing KDP crystals 74 cm long-lived radioactivation products in diameter is out of the question; associated with other commonly used it takes months to grow 27-cm-diam structural metals. Aluminum has only a crystals, and 74-cm-diam crystals would take years. Our solution was to assemble slices cut from 27-cm-diam crystals into a 3-by-3 mosaic array with a 74-cm-diam aperture The growth, cutting, alignment, final finishing, and crystal coating technologies were all either scaled up dramatically to handle the volume or were newly invented. Two of the most significant innovations are the use of a diamond cutting tool on a precision lathe to cut each KDP crystal surface to its final smoothness, its correct angle with crystal axes, and its required flatness. A second important innovation is the sol-gel that is used to provide an antireflection coating on the crystals, raising their output by 12%. The sol-gel is applied as a thin liquid layer that dries to form the coating. Without the development of these finishing and coating technologies, it would have been impossible meet the Nova cost, performance, or schedule goals.

l he Target System The Nova target system is Tdesigned to accommodate a wide variety of experimental programs. Tt Fig. 9 provides a variety of target-focusing and beam-diagnostic options at each of three Output of the Nova-like laser chain at (a) the fundamental wavelength wavelengths in two target chambers, it (1,05/im) and (b) the first harmonic is designed to operate with moderate (0.53 fjffl).

15 single isotope with a very low cross to develop many new approaches to the section for neutron capture, and the problems of isolation, amplification, resulting radioactive isotope has a half- stray-beam suppression, control, and life of only 2.3 minutes. Hence, (he only targeting. Construction was completed in lingering radioactivity will come from December 1984, on schedule and within impurities and alloying elements and will the budget of $176 million. Compared to be at almost unmeasurable levels, far previous LL\1. lasers. Nova is four times below minimum laboratory and more cost effective (measured on a government standards. joules-per-dollar basis) in constant Figure 10 shows the target chamber dollars. During Mova's activation period, installed in the Nova system, with the which extends into the spring of 1985, five west beams equally spaced around the laser will be tuned to meet the west axis. The five symmetrically performance milestones (most oi which placed east beams, entering from the were demonstrated with the Novette other side, are positioned so that each system). Weapon-physics and ICF aims between the west-beam ports. This experiments will begin in the summer of arrangement prevents one beam from 1985. The laser innovations and firing into the other. The five-beam inventions demonstrated here are already overlap spot at the common focus is less being extended to meet the needs for than 250 ^im in diameter, including larger, more efficient, higher-average- allowances for alignment, positioning, piuver laser systems.'" 15 and verification tolerances.

ummary The Nova laser system, although Sit incorporates many techniques and devices developed for the Argus and

Shiva lasers, surpasses all previous lasers Key Words: Aniiix trttjm-ncv tonvtT'.iiin. iCI; in power and wavelength flexibility. Its menial aMiliru-mi'TV fuMnn: KDC: MAI APItor*: advanced technology made it necessary \(i*.iru>:: rLi^rn.t ..huIUT: Shi\,v

Fig. 10 Ths Nova target chamber, a massive aluminum sphere 4.6 m in diameter with wall* almost 13 cm thick. The large flanges carry the Irequency-convereion arrays and the linal-locusing lenses with their positioning mechanisms.

16 Notes and References ? W \\' Simmons. /. T. Hunt, and W E 1. |. I.. Emmctl. [ H Nuckolls, jnd I E. Wood. Warren. "Light Propagation Thrmit;h Lar^e 'Fusion Power bv Laser Implosion.'1 5ti- Laser Systems,' IEEE I Qu.mtum Etvctnm Amn.. 230. 24 (1<)7-»). QE-17.'l72"(]98l). 2. D R Speck vt a/. "1.nr 5hivo Llser Fusion 8. I F. Hotzrichter, D, Eirnerl. E. V George, J. B Kjcililv. IEEE I. Quantum Electron. QE-I7. Trenhnlme. W IV Simmons, and J T Hunt. I5W(W]> Phytic* of Liser Histnn. Vol III High-Poiivr J. t. H. N'uckolk The Feasibility of Inertial Pulwd Lwr*. Lawrence l.tvermore National Cnnnnemenl Fusion, Ph\<. Tt\i.i\ 35, 24 Laboratory. Rept LCRL-52868, Rev, J E19R2) 0^2). *J M. A. Summers et at., A Two-Color 4. U' U". Simmons.. Engineering Design. Conf- Frequency Conversion System for Hij»h Power Sl HMO 11982). Lasers," IEEE/OSA Conf. on Laser 5, D L Kuiri'riRj. "Oscillator Develnpment for Engineering and Optics iCLEO). Tvch. Dig., 30 Ihe Nd:Glass l-.iser fusion Systems," IEEE, !. (|une 1981). Qtuntvm L'tvctnm. QEM7. 1694 (1981)- Ill I 1. Emmett, W i". Vpupki*. and J. B. h. G. Leppelmeier and W. W Simmons. Trenhnlme The future Dewtopinent of High- Scnw.iFUuu/ Koport, Lawrence l.uermore Power Solid Stale /jser Systems. Lawrence Xatinnal I.aburaiarv. Kept UCRI.-50021-73-1. Lieermnre National Laboratory. Rept. p 7fi. and UCKI. ^0021 73 2. p. in (1973) UCRl.-fi3344[1982). Building Nova: Industry Relations and Technology Transfer

A continuing technology transfer effort has characterized the Laboratory's development of progressively more powerful laser systems. The recent completion of the Nova laser is an example of a successful, long-term, major collaboration between Laboratory researchers and industry that has advanced the frontiers of laser technology and produced benefits far beyond the original expectations.

For further information contact Since the early 1970s, when laser affordable prices. The success of the Robert O. Godwin (415) 422-5H8. research and development were taking Laboratory's laser systems, from the shape at LLNL, technology transfer has early Janus and Argus lasers to the more been an integral part of the Laboratory's recent Shiva and Nova lasers, attests to laser program. D\ ring these years, the the productiveness of a close interaction needs of a burgeoning laser technology between research facilities and industry. steadily pushed forward state-of-the-art Although technology transfer has boundaries and placed stringent been a long-standing practice at the demands on an industrial complex Laboratory, it was formalized by not yet ready to meet the exacting Congress for all national laboratories specifications of laser designers. Our with the passing of the Stevenson- laser program made an active VVydler Technology Innovation Act of commitment to work hand in hand with 1980. The spirit of the Act is to increase Li.5. industry, ar 4 this cooperation of industrial innovation and to stimulate the researcher and manufacturer, born of economy of the United States by the necessity, has matured into a symbiotic transfer of technology from federally arrangement with benefits far exceeding funded national laboratories to domestic the original expectations. industry. Emphasis is not on a net Now, as then, this collaboration is a outflow of technology from laboratories two-way exchange: a means of passing to industry but rather on an exchange of to U.S. industry a developing piece innovative practices and ideas At l.LNL, of technology and a means by which our Technology Transfer and Exchange industry can give the research Office facilitates and monitors technology community high-quality components at transfer activity between industry and

18 the Laboratory and reports to the company must have a sufficient market Department of Energy, In addition, a to justify its efforts and a commitment to Technology Transfer and Exchange keep the technology up to date. A good Policy Committee comprising associate working relationship between the two directors, senior managers, and organizations is essential for success. representatives from various programs Our first step when making such a establishes Laboratory policy on transfer is to present a formal technology technology transfer and related issues. transfer symposium. We provide legal For the Nova laser system, about 75% notification in the Commerce Business of the total funds spent went to industry. Dailv and notify all interested companies Contracts for more than SI million were with whom we have dealt in the past. awarded to 25 companies, for $100 000 to Our first such meeting was in 1975 and SI million to 79 companies, for S10 000 to was associated with the Shiva laser SIODOOD to 224 companies; more than system. In the past few years, ive have 977 companies received contracts for less carried out a number of complete than S10 000. technology transfers (see Table 1). The interaction between LLNL and One example will illustrate the the industries that took part in building difficulties in effectively transferring the Nova laser illustrates the three basic and using a technology, even when the ways by which we transfer technology. receiving organization is very skilled and The first way is to tiansfer an entire qualified. When Bell Laboratories, an package of several diverse technologies outstanding technical organization, to industry; the second is to provide desired to duplicate our laser oscillator industry with specific technical for their research program, it took knowledge in a fairly narrow' area, and them a year to fully absorb this the third is to join with industry to technology: "We received more than develop a technology. In each case, we 250 drawings from Livermore for the may use any or all of the following oscillator and glass amplifiers alone. We methods: ordered over 2000 separate parts, had our • We work direct!-' with suppliers; for machine shop fabricate literally hundreds the Nova project, we had research, of precision mechanical parts for over development, and production contracts 200 drawings and then spent months with companies. assembling them. Altogether, we made • We have specific technology transfer four trips to Livermore, two trips to meetings, relating, for example, to power Stanford, spent two months ordering conditioning, coating damage, KDP parts and 2.5 man-years building the finishing, and plasma shutters system" (Ref. 1). • We provide limited consulting in the Other organizations interested in form of meetings, tutorial sessions, and building such a system can look forward specific consulting tasks. to considerable cooperation from the • W

Technology Recipient Comment

Shiva User system ILC Technology. Inc. h>rmal technology lransf:*r plus monv other* meeting li C has s.i|es of -S10M Kised upon rhis Trrhrnilrv^v

Short pulse oscillators ILC Technology Inc. Stpvi*r.il generations of Svkaniii. University of df'vt-kipmrnr SJIE-S hv II.C RocheMer, Bell laboratories, SI.AC. AWRE. China

Planar Iriodes PuKe Sciences Enterprises Cornp(inv formed on the basis of this technology

Praxis BBN, System Cognition. SvMfm Cognition formed on K'mii Ijh. Lockheed. the basis of this technology. Science Applications language bein^ used by others several invited papers

Absorbing glass calorimeters Apollo l-.iser. Scienrech Commercial product

Laser system architecture Mav.il Research l.abnratnrv. Linear chains with spatial filters KM5. Rutherford

Streak cameras Cord in Produced commercial product

Laser glass handbook Manv Standard reference jn opncal mdustrv

Shiva pulsed power Mjnv ffrrnai trchntrh^v trarr-ttT

Table 2 Examples of the transfer of specific LLNL technology.

Technology Recipient Comment

Electropolish cleaning Alliance T

Welding space frame Ri^me, International LHVI on Shi^a

Clean-room techniques Manv It \i\ - jnd 1^ short IOUTM

Vacuum technology Manv SE'\ eral Inrnjl meetings

Silver plating of reflector Manv Being used t»n \nvj

Diamond turning of KDF Many |r pro* r-ss (oi \o\ ,i

High-pressure freon cleaning Manv L*-ed on Shiva Vnettr. and \II\J

Ring electro-Pockels cells C li'vrland OvVaK \<>;v standard inmrrercul l.aM'rmelnis. .md Inrad prc*iUf«'

Standards for optical material Mans R^omnK .wvpted -urdard

Vibratory slress relief Hetmht L«-i'd on Shiv j

Welding with minimum vV-ldnn Ceneral Tool lit'ini; UM-ji on \ova distortion

20 INERTIAL FUSION electropolishing parts, cleaning parts, the initial light pulse is amplified Fig. 1 space-frame welding, clean-room to a 74-cm-diam beam and then focused Size comparison of KDP crystals from technology, vacuum technology, silver down to 250 ^m, such factors as high or the apple-sized crystal on the right to a plating, diamond turning of KDP, and low reflectivity, damage resistance, and boule large enough to yield a 27-cm' small specific technology transfer bits, optical transmission assume prime crystal for Nova frequency doubling. shown in Table 2. In general, this importance for all optical components. This demonstrates remarkable ad­ transfer has been to our suppliers, hut The goal is maximum energy delivered vancement in the making of solution- dean-room and vacuum technologies to the target. Therefore, the numerous grown KDP crystals. Until a lew years ago, the largest KDP crystals were have been transferred to many and large optical materials must be about 2.5 cm in diameter. We worked organizations. both of high quality and relatively with manufacturers 1o increase the size inexpensive. of crystal boules and to determine opti­ We worked with industry to produce mum potential sues for crystal arrays eveloping Technology for Nova's 74-cm oulput apertures. One Jointly large optical materials that would be of factor limiting the availability of large D The most beneficial and easiest the same high quality as smaller ones. crystals is their slow growth rate: 1 to transfer takes place when the Laboratory For this, we required three types of 2 mm per day. For instance, a crystal and a company work together, sharing optical material: BK-7, a high-quality boule 30 cm wide on a side takes 9 to the costs, to develop a technology. horosilicate glass similar to that used in 12 months to grow a rale of growth that represents both a risk and a cost liabil­ Generally, we either sponsor the research eyeglasses; fused silica, a very high- ity. Our goal was a system with the po­ and development within the company or quality optical material; and neodymium- tential of increasing the growth rate of work with the company in association doped laser glass. Schott Optical and KDP by a factor of two to four. with prototype or production orders. Such jointly developed technology is of long-lasting national beneiit because it results in significantly enhanced capabilities for the company. The following examples illustrate this.

Large KDP Crystals With the Nova laser, we convert the initial frequencv of the laser light to shorter wavelengths by means of harmonic conversion, achieved with large, pure crystals of KDP (potassium dihydrogen phosphate). We worked with three companies (Cleveland Crystals, Inrad, and Lasermetrics) to advance the state of the art of crystal growing from the production of small, apple-sized crystals to the growth of a KDP crystal weighing several hundred pounds and standing about a metre tall (Fig. 1). We procured these crystals from two of the suppliers, Inrad and Cleveland Crystals, for use on the Nova laser system, (See Ref. 2 for additional information about harmonic conversion.)

High-Quality Optical Materials The successful operation of any laser depends on the accurate reflection or transmission of the light beam, and this is highly dependent on the optical components of the laser—the lenses and mirrors that direct, transmit, or focus the beam. For the Nova laser, in which

21 Hoya Optics supplied the BK-7 glass. large optics pieces a metre or so in Becavse we bought about 45 pieces of diameter. Conventional polishing glass of the size shown in Fig. 2, low techniques for such large pieces would price was important. We obtained the have be™ very costly. Responding to second optical material, fused silica (used this need, two companies, Zygo to transmit the Nova third harmonic, Corporation and Eastman Kodak, built 0.35-//rn light), from Coming Glass and large flat-lapping machines similar to from Heraeus. Again, we worked with those used for polishing smaller units. both companies to ensure the deliv&y of The resulting scaling up of technology the required large sizes at high quality produced high-quality parts in larger and reasonable price. We procured the sizes at lower prices or, as needed, third material, neodymium-doped laser numerous small pieces. Figure 3 shows glass, from two companies, Schott the Zygo and Kodak flat-lapping Optical and Hoya: the latter has opened facilities. The Laboratory paid for the a plant in Fremont, California, to machines, and the companies designed produce laser glass and other precision the machines and provided all the optical materials. support facilities. The Laboratory's needs have top priority for use of the machines, but commercial uses afso are Optics Polishing possible. For example, Kodak also makes Another example of collaboration optical parts for copiers with its flat- lapping machine. involved the flat-lapping or polishing of

Fig. 2 Rough-cut BK-7 turning mirror blanks (1.2-mdram) manufactured by Schorl Optical. This company's ad­ vances in the manufacture of massive metis of BK-7 glass have played a significant role in making Nova a cost-effective laser. For eiample. Nova includes 98 massive minors, some of which have very high homogeneity to transmit the beam to diagnostics instruments. In addition. Nova laser chains tnclude two BK-7 lens blanks (the output spatial-fitter lens and the diagnostic objective lens) with a clear aperture of 77 cm. The borosilicate for all these components came from one continuous mefl, Schott Optical also developed apodizers that are made by hising a cylindrical section of index-matched, copper-doped BK-7 glass into a clear BK-7 substrate.

22 INERTIAL FUSION

Optical Coatings coatings. To obtain them, we paid for the The reflectivity level of an optical construction of large coating chambers at piece is determined by the type of both Optical Coating Laboratories, Inc. coating applied to it. Nova, being a large (OCLI) and Spectra-Physics (see Fig. 4). optical system, requires special optical The Laboratory paid for the design and

=ig.3 In 1981, Zygo Corporation and Eastman Kodak completed construction of large flat-lapping machines for finishing the large Nova mirrors: (a) Zygo's 4.1-m-diam machine, and (b) Kodak's 3 6-m-diam machine. Finished mirrcrs are produced at a combined rate of six per month. This is a significant accom­ plishment, in view of the size, of these mirrors, which range from 60.4 to 109 cm in diameter and weigh from 55 to 446 kg. All are high-precision mirrors hiving reflected wave-front distortions /./10 over the clear apertures and trans­ mitted wavefront distortions //B. Each machine features a Fizeau interferom­ eter with an 80-cm aperture for the fre­ quent checking of parts during the manufacturing process.

23 fabrication of the chambers, and the co.itings to meet commercial and companies paid for installation and Department of Defense requirements. housing. Laboratory needs have the These two chambers mark a significant first priority, but the companies also advance in precision-coating capability use these chambers to make optical in the U.S.

(b)

Fig. 4 (a) Nova coating chamber with a 3-m diameter at Optical Coating Labora­ tories Inc., can coat optical mate­ rials lor either high-rellecting or antireflecting properties, (b) The Spectra-Physics coaling chamber is cryogerwcaly pumped, a unique feature in the coating industry. The coatings are difficult to produce because of the tight tolerance (1%) required for the film thickness. Both companies have been able to produce polarizer coatings on large substrates, 10 Nova specification!, with eiceKent produc­ tion yields. Each has also produced a variety ol mirror coatings on substrates ranging from 60.4 to M cm in diameter- Five Nova coating designs have been produced, and three different sizes of mirror substrates (some of which are transmitting, for diagnostic purposes) have been successfully coated.

24 In addition, we assisted in developing, collaboration aimed at improving and for both the BK-7 and fused silica optics, extending existing technologies. The antireflecting coatings that are not easily result of this technology transfer damaged by laser beams. Schott Optical effort was that the required hardware developed (and we licensed) a neutral- was provided on schedule and within solution process to produce antireflecting specifications, and there was a surfaces on large BK-7 lenses. LLNL, concomitant significant increase in the Corning, and Wesringhouse developed a technology and production capability of coating technique called solution gelatin U.S. industry. U (sol-gel) to provide a similar low- reflectivity coating for fused silica. Key Words: laser—Argus, (anus. Nova, 5hiva; Stevenson-Wydler Technology Innovation Act ol ummary 19R0; technology transfer. LLNL's technology transfer effort Sduring the construction of the Nova laser system was motivated by our Notes and References desire to acquire at affordable prices 1. R. H. Storz. "Technology Transfer: A Case state-of-the-art components for our large History," User focus (September 1981). pp. 65-66. laser systems. Our exacting requirements 2. "User Fusion with Green and Blue Light,' made demands on the capabilities of Energy and Technology Ren'eii- (UCRt--52000.. our suppliers that necessitated a close 82-8). August 1982. pp.' 18-29. Managing the Nova Laser Project

The Laboratory's Nova laser, the most powerful in the world, is now operational. Effective management minimized the impact on costs and schedules of fluctuating levels of construction funding and coordinated the contributions of many subcontractors, both large and small.

For further information contact Experiments with the Laboratory's • 200 rrr of optical-quality surfaces. Robert O. Godwin <-l15l422-554S. Nova High-Energy Laser Facility will • 100 nr of optical thin-film coating. begin in February 1985. The Nova laser (For details of Nova technology, see the is the latest in LLNL's evolutionary series article beginning on p. 8 of this issue.) of large neodymium-glass research lasers. The Nova facility consists of two conventional structures, a 5500-nr office istory of the Nova Laser building and a 10 7D0-m" laboratory Over the past several years, buildin? plus the large laser system. H LLNL has built and operated The ten-beam, neodymium-glass laser a series of increasingly powerful and operates at three wavelengths, delivering energetic laser systems. (See the article in a 3-ns pulse 80 to 120 kj at 1.05 ^m on p. 1 for a perspective on neodymium- (infrared), 50 to 80 kj at 0.525 fim (green), glass laser research and development at or 40 to 70 kj at 0.35 pm (blue). Each of LLNL) Each of these laser systems was Nova's ten amplifier chains is 137 m more powerful than its predecessor by a long; the fundamental wavelength factor of five to ten. Shiva, for example, (1 05 /itn) is harmonically converted to was a 10-kJ laser, and Nova is a 100-kJ green or blue wavelengths bv unique system. In constant dollars, the cost of arrays of potassium dihydrogen each laser system scales roughly as the phosphate KDP crystals just before each square root of its performance. beam enters the massive target chamber The idea of the Nova laser was bom {fig. 1). Nova's electrical system includes in 1975. about midwav through the 10 000 capacitors, 5000 flashlamps, and construction of its immediate 110000 m of high-voltage electrical cable. predecessor, the Shiva laser. Nova was, Its optical systems comprise 1000 major in fact, originally conceived as an components, incorporating: upgrade of Shiva; in some of the earlv • 2000 litres of laser glass. documentation, it is referred lo as Shiva" • 1000 litres of fused silica. ("Shiva Star"), .During the conceptual • 10 000 litres of crown glass. studv for Novp, however, we decided • 150 litres of crvstals. that the most effective wav of exploiting

26 JNERTIAL FUSION

what we had learned from Fhiva would at a single frequency in the near- be to scale up the technology by an infrared.) Figure 2 shows the history of order of magnitude. The result clearly the Nova project schedule and would be more impressive than a star; milestones. hence, "Nova." (for a while, it was When the Shiva laser was dismantled known as Shiva-Nova.) in 1981, the Laboratory faced a hiatus in Funding of the Nova project was large-laser experiments until Nova would preceded by two years of conceptual become operational in 1984 or 1985. To planning and development funding. bridge the gap, and to provide a test bed In fiscal year 1978, Congress authorized for Nova components, we decided to the project at S195 million; it will be assemble an interim laser system. This completed at $176 million. Because facility, named Novette, was built in just of its size, the Nova laser facility is a 13 months and incorporated the first two congressionally funded line item—that is, Nova amplifier chains. This proved to a specifically identified construction item be a prudent decision. In addition to in the Federal budget. Although the enabling early verification of Nova project was authorized by Congress at hardware, Novette's experiments have the outset, funds had to be appropriated exceeded our expectations from a physics annually. This led to differences between standpoint, Among its achievements funds requested by LLNL and actual were a demonstration of wavelength funding levels. As a result of the variable scaling of fusion reactions and, of at least funding levels and some modifications of equal importance, the first confirmed the projects technical scope, it has gone demonstration of lasing at soft-x-ray through a number of cost and schedule wavelengths in a laboratory facility. changes. (For example. Nova's original Novette was operated for 18 months design called for twenty beams operating before it was dismantled last year for

Fig. 1 Target chamber of the Nova laser. Nova's ten beams converge to heal and compress a liny deuterium-trittum-fjlJed fuel capsule.

27 incorporation into Nova, having served description. For a project of the scale its purpose most admirably. of Nova, the key to structuring a manageable baseline is to break down ova Project Management the technical description into smaller and Under Department of Energy smaller pieces. The result is known as a N(DOE) guidelines, a project work-breakdown structure (WB5). The budgeted for more than about WBS serves as a baseline against which $50 million is classified as a major system cost and schedule estimates are prepared. or a major system acquisition. This In the Nova project, tasks were brings it under DOE regulations, which broken down into six levels (Fig. 3). spell out specific project management Level 0 is the Nova project as a single- techniques, milestone requirements, and unit undertaking. Level 1 consists of the reporting procedures. (The only other Nova laser itse'f and its two buildings Laboratory project subject to such (only the laser system is illustrated in the Fig. 2 controls is the Magnetic Fusion Test truncated example of Fig. 3). The next Nova project cost and schedule trends. Facility-B, scheduled for completion in level, level 2, comprises laser systems Construction was authorized in liscal fiscal year 1986.) such as mechanical systems. We proceed year 1978. The eleven events show the The first step in organizing a complex through subsystems and components to effect on project cost and schedule ot funding limitations, changes in project technology project is to establish a the lowest level, parts of components scope, etc. technical baseline or technical (level 5). The 31.5-cm-diam disk

Cost trend Schedule trend 250 -

5 6 £200

2 o

7 st , million s o f dol l 4

I100 | « 2 p so

I 1 1 1 1 1 1 1 1 1977 1977 1978 1979 1980 19B1 1982 1983 1984 1985 1986 1987 1977 1978 1979 1980 1981 1982 1S83 1984 1985 1986 1987 Date of change Date of change

1 Schedule slip due to fiscal year 1979 funding limitation. 7 Cost increase and schedule slip due lo "mark time" strategy and uncertainties about fiscal year 1982 funding and project scope. 2 DOE review approves ten beams only. 8 Cost increase and schedule slip for addition ot harmonic 3 Schedule slip based on reassessment of ten-beam-only strategy. conversion capability. 4 Cost increase and schedule slip due to fiscal year 1981 funding 9 Cost increase and schedule slip due to proposed fiscal year 1983 limitation. funding. 5 Preliminary DOE approval ot twenty beams and harmonic 10 Cost decrease and schedule acceleration due to full funding in conversion. fiscal year 1983. 6 Cancelled DOE twenty-beam approval and reverted back to ten 11 Schedule delay for extended Novetle operations. beams.

26 component shown in Fig. 3, for example, Laboratory's Engineering Department is composed of a housing, shield glass, a was already structured according to fluid pump, cradles and pipes, nitrogen personnel skills, we elected to carry over gas, and so on. that structure to the Nova VVBS At the outset of a project, it is not Having produced a technical feasible to specify a technical description description of the project and a WBS, our to the lowest level for every system; next step was to establish a one-for-one consequently the VVBS should be overlay for the project organization, so regarded as a tree that grows and that an identified person was responsible branches during the lifetime of the for each item at each level of the WBS. project. That person then prepared a technical Once the Nova project was defined in description, a cost estimate, and a terms of a VVBS, the next step was to schedule for his particular WBS establish an organization, or project responsibility. team, to accomplish the task. Such an Finally, to satisfy the DOE's reporting organization can be structured for requirements, we needed some means of management purposes in several ways. measuring scheduled performance One way is to match a particular group against actual performance. Such a with a hardware end item, such as an performance measuring system, called amplifier. Another way is to organize a cost and schedule control system, tasks according to personnel skills, such consists of three curves. The first two are as mechanical engineering, electrical conventional: an estimated cost curve engineering, and so on. Because the and an actual cost curve (Fig. 4). The

Level 0 Level 1 Level 2 Level 3 Level 4 Level 5 {project* (facility) (system) (subsystem) (component) (part)

Nova Laser Mechanical systems Amp* Bars Rod, 5-ennKpm Housing, ahMdglMi, fluid pomp, cradles and pipes, nitrogen gat, ate.

Dfak, 31.5-cm-diam

Disk, 46-cm-dlam

Fig. 3 Typical Nova project work-breakdown structure (WBS), shown here for only a portion ol the laser system itself. (Other leuel-1 facilities are office and labora­ tory buildings.) Such a detailed descrip­ tion provides a technical basis for man­ agement control. Because not every system can be initially described down lo its lowest level, the WBS grows in complexity and depth as the project is defined in greater detail. Roster of Nova Vendors

Contracts awarded for more than SI million

Ftavmond Kaiser Enginccr-i Cleveland Crystals I>igHat Equipment Corporation ScnH Cn. of California Eastman Kodak Company Wei dun International Sennit Class Technology Inc. DeJmonie Electrir Company Int. General Electric Co.. Capacitor IJiv. Ilnyj Op I its Incorporated Spectfj'Physics Inc. Alliance Tool & Die Corporation I Jnd K Builders General Contractor C. Overaa Ac Company Laivson Mechanical Conltactors Kictrit Pacific Company Ferre/o Electric Co. Riverside SleeJ Construction

Contracts awarded for $100,000 to SI million

Ovr^y/Pjcitic Manufacturing Ramtek Corp. Allied Engineering & Produclion Corp. I'npumo Precision Inc. Meyer Tool ie Manufacturing Co. Allen BrJdJy Lid. Bendii hit-Id Engineering Corporation Grinnrll Fire Protection Systems Co. Enterprise Roofing Service Inc Aydin Cnrpcratton Christie Constructors Tnc. Glass Fabricator; Inc. Chicago Bridge k Iron Company Pen hall Company Angenieux Incorporated Tm-,(ey Ijboratory Inf. Aero-Space Welding 4 Machining Inc. Hugin Industries IK Technology Inc. OK ford General Industries Inc. Floating Point Systems Inc. WeMinghause Electric Corporation Owens Illinois Inc. Kigre Incorporated Nump CJh-Hove Inc. H C and C Inc. The Hughes Plastic Co. Inc. Adam*, jnd Smith Inc. California Engineer Contractors Inc. Coleman Precision Mfg. Co- ».CI CTI Cryogenics U.S. Inc. CNR Corporation Ifellrethl Corporation |amcs L- Whit taker Inc. Bendix Corporation Srnilh Williston Incorporated Rest Too! Company limes Packaging Corporation Interactive Radiation Incorporated Hewlett Packard Flying Machines Electronics Metal Finishing Inc. CL Norton Co. Inc. Ken lab Mcjdvillr Precision Tool & Maiding Car Electro Forming Inc. RCA. I Ci. Smith Products Inc. Sunset Foundry Compjny Inc. Master Metal Products Co. Heraeuw\mersil Inc. Coors FAirceliin Company Red Feather Construct ton Inc. Molilor Industries Inr MadrLga Tron Woiks Pacific Electric Motor Inc. Lasrrmetrrc* Inc. Shelter Globe Corporation Coherent Inc. hr>lr pcranfk VfKijiirt Pan American Stee] Franklin High Voltage Inc. K linger Scientific C orporalion Quanfroni* Corporation Carat fnc. Sierra Crifle St Moist Co. Inc AM Weld Machine & Fabrication Co. Kinney Vacuum Co. Bechtel National Inc Reynolds Metals Co. Bach Machine Maxwell laboratories Inc. Kilsby-Boberls Company Adept Manufacturing Co. tut4 Construction Company Inc. Newport Contrail Corporation Aero Design Continental Optical Company Design Optics int.

Contracts awarded for $25,000 to $100,000

Zero Corporation Diamond Too) Company Mac Panel Company Dm Associates Abacus Electronics Company Inc. Stangenes Inc. A^lio Model JJcvelopmrnl Corporation Swcnlon Tool A Die Company inc. Kruffrl & Esser Co. I..T.I. C/O Prideau* Company Applied Optics Terminal Manufacturing Inc. MrMahon Welding Inc. Fisher Scientific Lukas Machine inc. Evergreen Industrie* United Detector Technology Chalet Tool Company Metal Hellows Corp. Champ Company Apsco Manufacturing Co. VA. Uarcy Company Atrtron Solid State & Optical Co. Deaurik XtMion Corporation DScM Machine Dresser Industries Caltinmetric Int Beimer Machine Works Kinttk System* March MeWlfab Inc. Edwards Thomas & Company Tapemalion A.I. Johnson Company Vers a lee lncr Lord Corp Hughson Metals IKO International BaUer* Corp. Lintech Corp. Harper-Leader Inc. Luick Quality Cage & Tool Inc. Acton Research Tcktrpni* McCarron Electric Math-Associates Snhio Engineering Materials Company Alpase Inc. Tom Walsh Associates ITiiT Electro Optical Products DJv. Cryogenic Experts Incorporated Wiegman & Rose- International Corp. I're^jy Corf Oracle Corp. t>aedal Inc Advance Engineering Plough Electric Supply Co. t'p'Righl Inc. (Juantel International Corporation Technical Fabricators Inc. Aracor lube Service Company Northeastern Tool Co. Inc. Cry^at Mark Micron Machinery Company Topaz Electronics Ahaus Tool Company Inc. AmrrJcorn Electronics Corp. Intrepid Enterprises Oriel Corf < ofHravtrs-(rfwrj" Corporation Integrated! Software Co. H-P Machinery < jrdon Instruments C o. Aladdin Healing Corp. FquiplD Inc. Aeroiech Inc. Analog Device Nuclear Data Inc. VAfjjn/FimJc Modern Plastics Inc. Pathway Bellows Crrttrti Tool Company Current Product* Inc. Fluorocarbon U.S. C/uarlz Division Arroioi Industrie* Inc. Western Piping Onan Corporation IUspan Precision Products Inc. Scientech Inc. Bearing Fngineering Krehner Manufacturing Compjny Inc.

30 INERTIAL FUSION

third curve, somewhat less familiar, is task and then arriving at a cumulative called the budgeted cost of work time to complete a budgeted item. By performed (BCVVP). comparing cumulative times with actual This last curve is the key to measuring labor costs on the Shiva project, we were a project's progress in meaningful terms. able to develop a composite labor rate. The idea can be illustrated with a simple Cost estimates were then made bv example. Suppose that vvc are building multiplying cumulative labor time bv cars and estimate that we can build ten iabor rates, which have been inflated for cars per month at a production cost of the year of expenditure. $10000 each. The projected total Together, the hardware and labor cost production cost per month would then estimates provided a basis for financial be $100 000. Let us further suppose that control of the Nova project bv enabling we know only that the total production us to compare, as described above, cost for a particular month is $100000. budgeted and actual costs of work We might be tempted to conclude that performed. we had precisely met our performance standard. However, we also would have Measuring Progress to know how many cars we had built at The progress of a large-scale that cost. If we had built 20 cars in a technology project can be compared to month for our cost of $100 000 the life cf an individual: it is born, grows (BCWP = $200000), we would have into childhood as it advances with performed very well, having produced learning, matures into peak productivity twice the scheduled number of cars for in adulthood, and finally enters a the same production cost, halving the terminal phase to make way for the next cost per unit. Suppose, however, that we generation of technology. Because, as in had built onlv five cars in a month for human endeavor, performance at each our $100 000'(BCWP = $50 000): in this stage of a project is judged by different case, we would have performed criteria, it is often necessary to shift dismally, producing only half the the management focus as the project number of scheduled cars at double the proceeds. The Nova project was divided unit cost. By enabling us to make such into three phases, with corresponding judgments about performance in terms of management emphasis appropriate to costs and schedules, then, the budgeted each phase. cost of work performed provides a mear-ingful measure of progress toward In the early phase of the project, a goal in time. characterized by design and procurement efforts, we focused on determining whether our purchase orders were being Estimating Costs placed on time and whether actual costs The utility of such a technique were no greater than budgeted costs. depends, of course, on accurate cost If so, we knew that two things were estimates at each level of the WBS. In occurring we were completing our Fig. 4 practice, this entails a continual refining The Nova project's financial planning of cost estimates for each element in the and tracking system enabled manage­ WBS as the technical description grows ment to measure progress against pre­ cisely specified schedule and cost in detail and complexity. For the Nova goals. In this illustration, two 01 the Estimated cost of work scheduled project, our initial hardware cost curves, the estimated and actual costs estimates went to a depth of level 4 or 5, (upper and lower curves) at a given that is, lo the component or part level. Budgeted cost of point in the schedule, are conventional We used the cost history of the Shiva t work performed performance measurements. To accu­ rately measure progress, however, we laser as a starting point, scaling the Actual cost of must know not only the cost of the work work performed -j Si costs up or down for larger or smaller performed but also the amount of work. components as appropriate. We added an This is represented by the third curve, amount for inflation and factored in a ths budgeted cost of work performed. learning curve for quantitv production. What appears to be good news about lower than expected costs can actually j Behind schedule We analyzed labor costs by first be bad news about performance if the estimating the time to complete a specific project Is behind schedule.

31 designs on schedule and our costs would companies with whom we contracted to be within budget. provide on-site design and operational During the second phase, as orders support for the laser system: Raymond began to arrive, we examined how costs Kaiser Engineers, the Bechtel Group, Inc., were accruing for the purchase orders. and the Bendix Field Engineering We normally paid on deliver,'; thus, if Corporation. Kaiser and Bechte! provided costs were being accrued as projected, on-site, integrated design support, and we could conclude that orders were Bendix provided on-site operational being delivered as expected. support. Finally, after the bulk of the orders was delivered, we needed to shift to DOE Review another index of progress. We chose to Because of the scale of the Nova focus on the activation of the laser project, the SAN (San Francisco svstem. We asked such questions as operations) office of the DOE established whether the components of Nova's ten an on-site project office. There were amplifier chains were being assembled, weekly informal discussions of various installed, and aligned as projected, and aspects of the project with the DOE whether the beams were being activated project manager. In addition, we held a on schedule. monthly project review with the DOE Nova project manager. Finally, we Subcontractors described techgical progress and reported Altogether, 195 subcontractors were on cost and schedule matters in great awarded Nova project contracts worth detail at one- to two-day semiannual $25 000 or more; 25 were worth more reviews. The semiannual reviews were than $1 million. As with the Shiva attended by people from the DOE Office project, our practice was to contract out of Inertia! Fusion in Washington, D.C., all hardware fabrication and to use from the DOE SAN office, and from our contract labor rather than increase laser program. Such reviews were helpful Laboratory staff to perform the same to our own laser program management work on a short-term basis. This policy as well as to off-site personnel. minimized manpower fluctuations within the Laboratory. Most of the hardware onclusions was procured on a fixed-price, build-to- Capable and experienced print basis: we provided the design and C management is a necessary but contracted with a vendor to produce the not sufficient condition for the success of item according to our drawings rather a large-scale, high-technology project than to performance specifications. Such such as the Nova laser facility. The a policy enables a small, low-technology sufficient condition is a team of firm to produce a high-technology knowledgeable, dedicated technical staff component. Here again, a detailed WBS who will work in harmony to achieve proved invaluable; by separating a the project's goals. At LLNL, w _• are complex, high-technology system into fortunate, indeed, to have retained with many smaller elements, we enabled a relatively low turnover a large team of large number of small firms to participate highly skilled specialists who have and also minimized the risk for each worked together on high-technology vendor (and for the Laboratory, too, by laser systems for more than a decade. spreading the costs over many suppliers The effectiveness of such teamwork has rather than over a few large contractors). been demonstrated bv the successful All told, about 75% of the dollar value completion and operation of the Argus of the $176 million Nova contract was laser, the Shiva laser, the Novette laser, contracted out to private industry in the and now the Nova laser. L2 form of hardware procurement (see the roster of vendors with contracts worth more than $25,000, on p. 30). Key Words: huj>;*'lfi1 tti-.t ul ^rk hrfakLlnnn \mutiirt' Optical Coatings by the Sol-Gel Process

••''••...: i "'••.•rin^ '-.-:•' [{ie liifj,--'- : ;.:H'L! • •-> H 5 d ,• •!!'- .<-•:•'.: •;• ;;"•: \cv«i i.>.>-' j'Hi we tiro v:i Ini-.'ci';ir_; -Aitb -:- -^n';i!

phen light strikes a piece of the laser's optical elements of larger polished glass, and again diameter, thereby reducing the light Wwhen it leaves, most of it is energy per unit area, but this would transmitted but some of it is reflected. be far too expensive Another way The amount of reflected light and its would be to limit the power of the degree of polarization depend on the laser, again reducing the energy per angle of incidence. Even for normal unit area, but this would limit the incidence, however, the amount of kinds of experiment that could be reflected light can be as much as 5% conducted. Instead, we elected to per surface (both front and back). develop our own, highly damage- For many purposes, this resistant, antireflective surface transmission loss can be tolerated; treatments that can withstand the however, in a complex optical system power of the \ova laser. containing many elements, the losses One method that works well on add up. To minimize these losses, it lenses made of borosilicate glass is common to apply some sort of involves etching the glass surface to antireflective coating. Many tvpes produce a graded index of refraction.1 of antireflective coating have been It is the abrupt change from the index developed to improve the efficiency of refraction of air to that of glass that of optical systems. Most of these produces surface reflection. The function by producing two reflections porous surface layer provides a more of equal intensity that cancel each gradual transition that eliminates most other because one is 180° out of phase of the reflection. with the other. Not all of Novas optical elements When it comes to high-power could lie made of borosilicate glass, optical systems such as the Nova however Many of the experiments laser, however, none of the common planned with the Nova laser require antireflective coating materials is short-wavelength ultraviolet light, not suitable: they all suffer damage the fundamental infrared wavelength. because particulate impurities cause To convert the infrared light to shorter spatially localized absorption of the wavelengths, we pass it through an intense laser light. One wav to arrav of potassium dthydrngen compensate for this might be to make phosphate (KDP) crystals. However, upon exposure to ultraviolet light, additional layers to build up the borosilicate glass solarizes (turns coating to whatever thickness is brown and opaque). Therefore, for all required to match the desired optical optical components following the wavelength. frequency-conversion array, we must use fused silica, the only material he Sol-Gel Process presently capable of transmitting high- Our method for preparing intensity ultraviolet light. Tdamage-resistant antireflection Since fused silica has a different silica coatings is a particular composition than borosilicate glass, application of the sol-gel process," etching it does not produce a porous a very versatile process for making laver. Hence, we had to develop metal oxides that involves: another way to make a porous silica • Dissolving one or more metal- layer of controllable thickness. Silica, organic compounds in an organic itself, is verv damage resistant and solvent. thus would be an ideal material for an • Gelling the resulting solution. antireflection coating on fused silica. • Converting the gelled product, The sol-gel process we have usually by drying and/or heat, to an developed enables us to form, from oxide. solution, roughly spherical silica To make an oxide coating on a particles that are all approximately the substrate using the sol-gel process, for same size. By varying the reaction example, we can spread the solution conditions, we can van- the particle on the substrate and let the gelation size from about 8 to 1000 nm; we take place passively by solvent have found that the best size for our evaporation. The form of the resulting antireflection coatings is 20 nm. oxide, whether dense or porous, Although the silica in these particles depends on many factors, including is dense, the spaces between the the chemical makeup of the metal- particles make a layer of them quite organic compound used and the porous. The silica in the particles has amount of water in the initial solution. an index of refraction of 1.46, but the The general sol-gel process can be overall index of refraction of the used with many different metals to coating is about 1.22 because it is the form many different metal-organic average of the particles and the space compounds. To be suitable, the bettveen. derivative compound should be With the sol-gel process, we can readily convertible to the metal also apply antireflection coatings oxide or hydroxide, preferably by directly to the KDP crystals, which hydrolysis. An acceptable but less would be damaged bv exposure to the desirable alternative method of water or heat used in other coating conversion is by thermal degradation, processes. This greatly simplifies the Metal-organic compounds of the construction of the frequency- alkoxide class are particularly suitable conversion arrays, eliminating the for the sol-gel process, The generic need to sandwich the crystals between chemical formula of the alkoxides is: fused-silica windows and fill the assembly with index-matching fluid M(OC H,„..), to eliminate internal reflections. With n the antireflection coatings, the KDP where M is a metal, n is an integer, x crystals can be exposed directly to is the valence state of the metal, and tV laser beam. the other symbols are the standard The antireflection effect of such element abbreviations. Specific a laver is much greater at some examples of the alkoxides of different wavelengths of light than at others, metals include: and the wavelength of maximum BfOCjHJ,, Si(OC H,) . Ta(OC H,) effect depends on the thickness of the : 4 : ; coating. Fortunately, we can deposit AI(OC4l-U,, Ti(Ot,H7}., \b(OCH,)-, All of these alkoxides are liquids, product initially remains in solution. completely miscible with most organic The solution at this stage can be solvents and readily hydrolyzed to give applied to a substrate and converted the metal oxide and the parent alcohol. to an oxide as before. The titanium alkoxide noted above, for Alternatively, we can allow the example, combines with two molecules polvmerization to proceed to the gel of water to form one molecule of stage; Ihe solvent is retained but the titanium dioxide and four of propyl metal derivative is now insoluble in alcohol: the form of a crosslinked, three- dimensional structure. Subsequent

Ti(OC3H-)4 t 2H,0 — treatment of this gel can produce

TiO, + 4C3H;OH porous or dense oxide products. In some cases, the solution can One way to deposit an oxide assume yet another form: a discrete coating is simply to spread a solution second phase consisting of extremely containing one of these alkoxides small, colloidal oxide particles instead uniformly on a substrate. Upon of a gel. It is these colloidal solutions exposure to moist air, the solution will of oxide particles that we are using to take up all the water needed for the make antireflective coatings for the hydrolysis reaction shown above. If we fused silica and KDP optics in the Nova want to deposit a mixed-oxide coating, laser. we can dissolve two or more alkoxides in the same solvent. Another deposition method starts Jk ntireflective Coatings with the addition o/ a small quantity X^k Optical theory dictates that, of water to the metal-organic solution X> A> for optimum performance, to initiate hydrolysis. In many cases, a single layer of antireflective coating particularly with silicon alkoxides, for fused-silica optics have a refractive this leads to polymerization, and the index of about 1.22. The refractive index of silica is about 1.46. Therefore, if we want to make the antireflective coating of silica, we will need to achieve a porosity of about 55%. The SifOCjHs), - H,0 sol-gel process is particularly applicable under these conditions because il yields I silica of the highest purity {required to keep the laser damage threshold high), (CjH O) SiOH - CjH OH and it is easy to obtain the required s 3 5 base porositv. Fig. 1 acid (CjHsOfe Si(OH), In our investigation, our silica source Alternative reaction (CjHsOfe-Si-O-SiWCjHsfe is tetraethy! orthosilicate. Si(OC:,H,)4. pathways in the hy­ a volatile liquid that boils at a drolysis of tetraethyl Hoom temperature of 167°C and is readily orthosilicate to sili­ temperature con dioxide (silica). 80°C purified by fractional distillation, thus In the reaction using giving very pure silica. Hydrolysis an acid catalyst, the 5i(OH)4 requires a catalyst that ma> be either coating consists of acidic or basic. The pathway along a polymer that de­ OCjH5 grades to silicon di­ Room which the reaction proceeds depends oxide only at high -- Si—0-- temperature on the choice of catalyst, but the end temperature, some- product in either case is silica (see limes leaving behind OC,H 5 Fig. I). The overall reaction is: harmful organic resi­ SiO, dues. When the cata­ lyst is a base, the 4S0-C 5i(OC2H,)j - 2H:0 -

reaction proceeds SiO, - 4C:H,OH entirely at room tem­ perature and yields SiO, At first, we investigated the acid- pure silica. calalv/ed system. The first step is to

36 INEKTIAL FUSION

dissolve the ir'Melhyl orthosilicate in boiling ethanol and partially hydrolyze it to produce a soluble ethoxysiloxane polymer. We applied this polymer to the substrate and converted it to silica by thermal decomposition (baking at 400°Q. then further increased the porosity of the silira coating by etching it in a dilute solution of hydrofluoric acid. Although the optical properties of the resulting coatings were quite satisfactory, some of the coatings failed under irradiation with laser light. The cause of these failures probably was incomplete thermal degradation of the organic components in the orginal siloxane. This would leave carbonaceous residues that would absorb energy, causing destructive hot spots in the coatings. To avoid this source of trouble. we lumed our attention to the base- catalyzed system, which produces a colloidal dispersion of silica particles Fig. 2 in ethanol (Fig. 2). The preparation is Photomicrograph of the uniform colloidal particles of silica produced by the base-catalyzed sol-gel carried out at room temperature and process. The particle size can be controlled by adjusting the reaction conditions. The refractive takes two to three days. The silica index of an antirellective coating made of these particles is less than that of the silica particles particles are roughly spherical and can because it is averaged over the pores as well as the particles. be made in anv diameter from about 8 to 1000 nm by varying the reaction conditions. After we prepared the base sol with silica particles of the preferred size (about 20 nm). the coating is applied simply bv spreading the sol evenlv on the substrate and allowing the ethanol to evaporate at room temperature. This eliminates high- temperature baking and with it the possibility of unwanted organic residues. In Fig. 3. we compare transmission spectra of a 5-cm-diameler fused-silica disk at several stages in the process of coating it with three layers of the sol mixture. Two features of these spectra Fig. 3 aie immediately apparent: the narrow Two-surface transmission spectra of a piece bandwidth over which transmission of fused silica illustrating the effect of add­ rises to almost 100% (that is, near-zero ing successive layers of our antireflective reflection), which is characteristic for silica coating. Uncoated, the fused silica re­ this type of coating, and the shift in the flects about E% of the incident light over a broad range of frequencies. The first coating specific wavelength for zero reflection layer produces almost 100% transmission as we apply additional coats of silica. (zero reflection) but at a very short wave­ This latter property makes it possible length (220 nm). The second and third layers to tailor the coating for a desired successively broaden the range of zero re­ wavelength range. 200 400 600 800 flection and shift the peak to longer Wavelength, nm wavelengths.

37 Similarly in Fig. 4, we compare the have scaled il tip from laboratory size transmission spectra of a 5- by 5-cm to more than 750 litres, enough to coat KDP crystal before and after receiving the large Nova lenses. To obtain a one coat of silica sol. Since the light uniform coating on these lenses, we set changes to a shorter wavelength upon them on edge in a deep narrow tank passing through the KDP crystal, the filled with the coating sol and then antireflection coatings must be of drain out the sol at a constant rate different thicknesses on the two sides, (Fig. 5). Because this process produces thicker on the first side that sees the coatings of equal thickness on the two infrared light and thinner on the second sides, it is not suitable for coating the side that passes the converted, shorter- KDP crystal for the frequency- wavelength light. conversion array. Instead, we set the Table 1 presents typical damage- KDP crystal spinning on a turntable, threshold levels for coated fused- and pour the coating sol gently in the silica and KDP samples at various center. Centrical force spreads out the wavelengths, and compares them with liquid in a thin sheet, leaving a uniform the surface-damage thresholds of coating on one side of the crystal. After uncoated fused-silica samples and the enough layers have been applied to bulk damage level of KDP. The coated reach the desired thickness, we turn the fused-silica samples can withstand as crystal over and coat the other side. much laser light as can uncoated fused silica; the sol-gel coatings outperform ielectric Mirror conventional antireflective coatings by Coatings three- to fivefold. Furthermore, the DHaving learned how to coating on the KDP crystal can manipulate the index of refraction of a withstand more intense laser irradiation surface coating to suppress reflections, than can the interior of many KDP we can turn the process around to crystals. produce highly reflective coatings to We have now thoroughly established make the mirrors that turn and guide this base-catalyzed sol process and the laser beams. To do this, we coat the substrate with many alternating layers of a material with a high refractive index and one with a low index. As with the antireflection coating, the mirroring effect is specific for a particular wavelength range that depends on the thickness of the individual layers. Dielectric (nonmetallic) mirrors of this type are widely used in some laser applications but have been limited in other cases by their inability to withstand intense laser irradiation. There art' a number of oxides with high indices of refraction that are quite readily prepared from alkoxide starting materials bv the sol-gel method. Fig. 4 Among them are the oxides of titanium, zirconium, hafnium, tantalum, and Two-surface transmission spectra of a KDP crystal showing the effect o< our niobium. The intervening layers of low- antireflection coating. Uncoated, tha crystal refractive-index material are commonly reflects 8 to 9% of the incident light over a made of silica. broad range of frequencies. When coated The dielectric mirror effect depends with one antireflective layer, the crystal partly on the number of layers and transmits almost 100% (zero reflection) of the incident light for wavelengths near 400 600 BOO partly on the difference between the 380 nm. Wavefenglh, nm indices of refraction of the two

36 INERTIAL FUSION

materials. To minimize the number of layers and the expense and technical Table 1 Typical damage threshold levels for silica antircflcction coatings and optical difficulty of fabrication, the difference materials used in the Nova laser. must be as great as possible1. Hence, the high-index layers should be as dense as Laser Surface damage threshold possible so that they have the highest Pulse Silica COJilin g " vc fused Silica coating possible refractive index; this normally Wavelength, length. on fused silica, silica, on KDF, requires a high-temperature bake. The nm ns !/cm! 1/mv J/cm! silica layers, because they should have

the lowest practical refractive index, can 24£ 15 -1-5 fi-S _.. remain porous. 353 0.6 8.5-10 1(1 .J-5" We have been experimenting, so far, 10M 1.0 10-14 If. — with dielectric mirrors produced only on a small scale, mostly with layers of "Samples iml lesled at this wavelength. tantalum oxide and silica. For the ''Damage threshplii of [he coating was tint measured dirertly iH'iaust' hulk damage mairreil tantalum oxide layers, we have used first in tho-v samples :estisl. two alkoxide starting materials:

tantalum pentaethoxide, Ta(OC2H;}?, and tantalum sesquichloride elhoxide,

Ta(OC2H,),,CI:,. Either of these materials is applied to a substrate (usually fused silica) from solution and allowed to hydrolyze to form the oxide, as described earlier. The reactions are, respectively,

2Ta{OC\H0, + 5H.O — Tap," + 10C;H=OH , and 2Ta(OC,H-)^CI,, + 5FUO —

Tap," 4 5C,H5OH + 5HC1

We found that we could obtain coatings of high density with retractive indices of about 2.1 by baking the specimens at 150°C To make the silica layers, we use an acid-hvdrolyzed ethyl silicate solution instead of the base-hydrolyzed system used for the antireflection coatings. We apply this solution to the substrate and heat it to 1^0°C as with the tantalum oxide layers. On fused-silica substrates, we have been able to make durable dielectric mirrors thirteen layers thick, six layers of silica alternating with seven layers of tantalum oxide. Currently, mirrors iviii, more layers tend to crack and peel Fig. 5 during the baking operation, possibly One of the large (80-cm-diameter) tused-silica lenses for the Nova laser being lowered into the because of stresses generated by tank where it will receive its sol-gel antireflective coating. Once the lens is in place, the lank is differential thermal expansion. filled with about 750 litres of colloidal silica solution. After a wait of about 15 minutes to allow the atmosphere between the liquid surface and the l3nk lid to come to equilibrium and all waves and Figure 6 shows the transmission ripples in the solution to dissipate, the liquid is drained away at a constant rate. The rate at which curve of a nine-layer dielectric mirror the liquid is drained away controls the thickness of the coating; a fast rate produces a thick (four layers of silica alternating with coating, a slow rale a thin coating.

39 Fig. 6 incorporating them in many types of coating. With this process, we have Transmission spectrum of a dielectric mirror coating on one side of a fused-sMIca sub­ developed a superior method for strate. The coating consists of four layers of preparing highly damage-resistant porous silica (low index of refraction) alternat­ antireflection coatings of porous silica ing with five layers of dense tantalum oxkfe for the large fused-silica lenses and (high refractive index). The deep dip at about KDP frequency-conversion arrays used 400 nm indicates that 91% of the Kght of this wavelength was reflected. Almost all the fight in the Nova laser. We also have of wavelengths longer than 500 nm was trans­ produced small, experimental, damage- mitted. The drop-off at wavelengths shorter resistant dielectric mirrors of tantalum than 270 nm is caused by absorption in the and silicon oxides with a variation gf tantalum oxide layers. the sol-gel process. With this so!-gel method, we probably will be able to make damage-resistant antireflection coatings that can withstand lasers even more powerful than Nova. With further development, this sol-gel process can be extended to many other types of optica! coating. U

400 600 Wavelength, nm Key Words: a!ko\ide; coating—antiteilection. damage threshold, frequency conversion: laser— Nova; lens-^fused silica; light—ultractolet; mirror—dielecinc; potassium dihydrugen five layers of tantalum oxide, all on a phosphate (KDP); -ol-gi-l. silica disk). This mirror reflects about 91% of the light at a wavelength of 400 nm. Longer wavelengths pass straight through with little loss; the Notes and References tantalum oxide strongly absorbs 1. for a discussion of antireflective coatings wavelengths shorter than about 280 nm. for borosilicate ftlass optical component*, see the March 19JC issue ot Envrgy Ami ummary Tn-hiwhw KWwiv (L'CRI--53fl«l-is:-3). . 9. The sol-gel process is a versatile P The ol-cjel process u as described briefly new method for preparing the S in fc'neri^i- Jnd h\hnoh\tiv /cevieiv (UCKI.- oxides of manv different metals and =.2000 B4 7). Inly I9W. p. 43

40 Frequency Conversion of the Nova Laser Arrays of potassium dihydrogen phosphate (KDP) crystals are used to convert the infrared light of the Nova laser to shorter wavelengths in the visible and near- ultraviolet for improved target performance.

For further information contact he performance of inertial- coherent and virtually lossless. The Mark A. Summers 1419) 423-2861. confinement fusion (ICF) generation of harmonics in optically Ttargets is known to improve at nonlinear media is the best laser wavelengths shorter than 1 /jm, understood frequency-conversion for example near 0.5 or 0.3 jjm (visible technique, the one that provides most green or near-ultraviolet light). At control over beam quality, and the these shorter wavelengths, ICF targets one adopted at LLNL. absorb more light and the deleterious In the process of harmonic prehearing of the fusion fuel is generation, very intense laser light reduced. However, most high-power incident on a transparent, optically lasers considered for fusion nonlinear medium interacts with the applications operate either near 10/Jm material's atomic structure to generate (the carbon dioxide laser) or near electromagnetic radiation with 1 pm (Nld:glass lasers). Therefore, frequencies that are multiples of the to produce the desired shorter frequency of the incident light (see the wavelengths, we have two choices. box1 on p. 4). Such multiples are Either we can develop a completely known as harmonics of the new short-wavelength iaser, or we can fundamental frequency. These convert the fundamental frequency of harmonics are commonly designated an existing 1-pm laser using a as lot, 2d). 3to. etc., where \ is the third lasers has been pursued at LLNL and harmonic, etc. For Nd:glass lasers, elsewhere as the more reliable and \ia corresponds to a wavelength of cost-effective approach to obtaining 1.05 fivn, in the near-infrared; 2w shorter wavelengths. corresponds to a wavelength of In principle, any physical 0.527(im, green near the spectral peak phenomenon that changes the for visible light; and 3w corresponds frequency of light is a candidate for to 0.351 ^m, in the near-ultraviolet frequency conversion. In practice, just beyond the visible spectrum. however, since we also require a focusable output, a high conversion istory efficiency, and a high damage Frequency conversion of laser threshold, frequency conversion must H light was first demonstrated be based on a phenomenon that is in the 1960s, when experimenters

*'/4 2 passed the deep red light of a ruby By 1980, frequency conversion as laser through a quartz crystal, applied to ICF laser systems was well producing a harmonic in the established. The nonlinear material of ultraviolet range.2 At about the same choice, KDP, had been used to rime, other researchers demonstrated demonstrate conversion efficiencies conversion of the infrared output of greater than 50% on high-power laser an Nd:YAG laser to an ultraviolet systems around the world at aperture harmonic by illuminating a crystal diameters as large as 15 cm. The next of potassium dihydrogen phosphate challenge was to scale up the (KDP).3 frequency-conversion systems to the This frequency-conversion Nova size (74-cm aperture). capability gTeatly increased the Construction of the Nova flexibility of neodymium-based laser frequency-conversion system has media (e.g., Nd:YAG, Nd:YLF, and required considerable technology Nd:glass). In fact, the potential for development. Over the last five years, high performance at short the most significant achievements in wavelengths was one reason we the technology of large-aperture selected the Ndiglass laser system frequency conversion have been: over the much-longer-wavelength • A new geometry for KDP crystals carbon dioxide laser for 1CF research, that efficiently produces both 2(0 and since the benefits of target irradiation 3w light. with short wavelengths had been • Methods for fabricating large anticipated.^ plates of KDP crystals accurately Through the 1970s, frequency- aligned in relation to their conversion techniques were applied to crystallographic axes. Nd:glass laser systems in this country • Techniques for mounting the and abroad. The research team at KDP plates in large arrays. KMS Fusion, Inc. (Ann Arbor, • A high-damage-threshold Michigan), was the first to conduct antireflection coating for KDP. plasma physics experiments using the second harmonic (2io) of an Nd:glass DP Geometry laser. They used a KDP crystal with a The Nova frequency- clear aperture approximately 10 cm in conversion system is an 3 K diameter. At Ecole PoMechnique in extremely simple and .flexible device France, 1CF laser-plasma interaction that can efficiently produce 2o> or 3) and fourth () using KDP.' light, with a conversion efficiency of Here at Livermore, the Argus laser approximately 67%. A second KDP system was frequency-converted to crystal, oriented differently, mixes the study plasma conditions at 2m, 3CJ, 2CD light with residual lai light from and Aw!* With the Novette laser, the the first crystal to produce 3m light, advantage of ICF implosions using with an intrinsic conversion efficiency shorter wavelengths was clearly approaching 90%. demonstrated at the multikilojoule At Livermore. we recognized that a level, where, by irradiating targets modification of an LLE design would with visible (2(o) light, higher fuel efficiently and conveniently produce densities were achieved than had 2(D as well as 3tu light. Two KDP been passible with Shiva experiments crystals, optimized for 3) are oriented so that they produce two wavelength.^1 components af lea light, orthogonally Harmonic Generation

When a low-intensity light wave propagates through Laser pulses typically have a spectral brightness a transparent medium, it drives the electron cloud of the (power density per steradian-hertz) many orders of constituent atoms into oscillation (a). This oscillating magnitude greater than that of incoherent sources. It >s electric charge constitutes a polarization wave that gives this property that enables a laser pulse to produce a rise to a radiated wave. The radiated wave represents more vigorous oscillation of the electron cloud. The the usual "linear" response of the medium to the electrons may be displaced from their equilibrium incident light. Because the oscillatory motion of the position by a distance many times larger than the truly electrons follows the vibratory morion of the driving minute excursion induced at low intensity. As a result, field, the radiated wave is of the same frequency as the electrons come closer to neighboring atoms, probing the incident wave. The wave thus emerges from the their repulsive potential with each oscillatory cycle. This transparent medium with its frequency unchanged but process enables the structure of the medium to influence with a phase delay that depends on the material's the frequency of light radiated by the oscillating refractive index. This applies to all low-intensity light electrons. If the strength of the repulsion experienced by waves passing tnrough transparent media. an electron driven by intense laser light is asymmetric (different in opposite directions), its motion will be biased (b). If we decompose an electron's motion into its frequency components, they appear at multiples of the input frequency, la (c). For a centrosymmetric site, the (*) components appear at odd multiples (mostly 3a> and, to Atomic site Lattice repulsion a vastly smaller extent, 5(D). for an asymmetric site, we obtain both odd and even multiples. By far the strongest of these is 2a and, in order of rapidly decreasing strength, 3m, 4

- Electron motion

(b) <0

Symmetric repuWve potential Symmetric electron motion (oniric lito) '\/+

'Y^ Aiymmetric reomrtve potential '•»'\S\J~\J~\S\ V S (acentric atta) s-WWWVA.

44 INERT1AL FUSION

inKDP

outward at the phase velocity of the medium for that orient a crystal of KDP so that the propagation velocity frequency, which is the vacuum speed of light divided of incident light waves produces exact phase matching.

by the refractive index at 2(0 (v7a = c/nltll). The The optical axis of KDP is such that light polarized fundamental wave, however, propagates with a different perpendicular to the axis experiences an angle-

phase velocity (v,w = c/nlw}. Because the fundamental independent (ordinary) refractive index, n^ whereas wave imposes its phase on each of the electron light of the other polarization experiences an angle- oscillators, and thus on the radiated 2a> wavelets, the dependent (extraordinary) refractive index, n^ see

net radiated wave at 2o» is the sum of many out-of- figure (e). The extraordinary index ne(0) at the second phase wavelets from the host of oscillators throughout harmonic is equal to the ordinary index at the first the material. We describe this situation by saying that harmonic at a particular angle 9, the phase-matching the process is not phase matched (d). The result is a angle of incident light. This condition, called type-1 very weak conversion of the laser energy from the phase matching, produces phase matching for the fundamental to the desired frequency, 2(0. second harmonic. To achieve the phase-matched condition, in which Phase matching may also occur when the second the 2(0 wavelets will add coherently, we must arrange harmonic is matched with two fundamental waves for the phase velocity of the fundamental wave to equal polarized in orthogonal directions. This is called type-II that of the second-harmonic wave. This can be done by phase matching. In this way, a KDP crystal provides a choosing a material with anisotropic characteristics that high conversion efficiency in one particular orientation compensate for the variation in phase velocity with but a low efficiency in any other orientation. We have wavelength. The wavelets from all the oscillators in demonstrated experimentally this conversion technique, such a material then will reinforce one another (d), called angle-tuned harmonic generation, with the resulting in a high conversion efficiency. KDP crystals Laboratory's Argus and other laser facilities. This display this property. technique requires precise orientation of the KDP The phase velocity of light in a transparent medium crystal, typically to within 100/irad. is inversely proportional to its refractive index, which, in rum, depends on the spatial orientation of the material and on the wavelength of the light. It is possible to

(e) m WWV \\\ Not phiH matched /\ /\ /A f\ AA \\\ 'VWW \\V Input wave it 1w JZ, v * s—s '""*• s

>._• *..«.' •V--' Phaae matched 'VVVW o/wVW\Aw/ *2n Wavttength

45 polarized with respect to each other; crystals, with input and output surface "quadrature" is used to refer to this normals approximately parallel to the mode of operation. This design led to phase-matching direction. We found the two-crystal architecture, referred to that the input and output surfaces as Quadrature 2cu/Cascade 3300^rad). Nova beam can efficiently produce 2 to that the lack of orientation accuracy 3ft> conversion is accomplished by was a result of the x-ray diffraction small rotations in two axes, permitting measurement technique and a rapid changeover. This design thus inaccurate crystal machining. offers the great advantages of Fabricated in this manner, each Nova simplicity of operation and crystal would have to be individually commonality of parts between aligned to the Nova beam to obtain Fig. 1 optimized 2to and 3u> crystal arrays. high conversion efficiency. The Tho frequency-conversion array* are located complexity of providing individual >l the output of each Nova beam line, juit be- abrication of KDP alignment adjustments on such a large lore the target locuting lentei. Simple rota­ Crystals number of crystals made this a very tion adjustment* ol the array* provide Nova When we fust started working unattractive approach. with the nexioilrty to irradiate target* with F three wavelengtha of Kght infrared (1HJ), vial- with KDP crystals, we polished them Therefore, we took a different ble green (2(d), or near-uitraviolet (3w). using techniques similar to those approach to controlling the fabrication developed for glass polishing. Our of KDP crystals. Our object was to goal was to produce optically flat Ooubler improve the orientation accuracy so array /~ Mixer that alignment adjustments of Lena ^ / array individual crystals in an array were not required. Instead of using x-ray diffraction to determine the phase- matching direction, the crystal under I.OSjen 1.06/an test is used to frequency-convert a small infrared beam of the same wavelength as produced by the Nova laser. The orientation for peak conversion efficiency is measured by Doubter comparison with a standard crystal array ^ y-Mixer and the angular correction needed to Lena Rotate 45° \ / array properly locate the surface normal is determined. The next challenge was to 0L527*m VBB im accurately machine the surface of the KDP crystal, which is very soft (about the hardness of chalk). Therefore, we turned to the highly precise technology of single-point diamond machining. We successfully developed Doubter the methods and tooling for

M er Rotate 35" """* "^ f~ ™ Lens machining KDP to the exacting Till 1/*° 'VA array angular tolerances required for Nova. In fact, this machining technique is so precise that it can produce the final 0.361 **n 1JC jen surface finish required (

46 INERTIAL FUSION

Several crystal plates are cut from a . process was extremely time Fig. 2 single boule (Fig. 2) using a "water- consuming and costly; finishing Large KDP crystal boule (grown by Cleveland and-string saw" (KDP dissolves in accounted for approximately 50% of Crystals, Inc.) will yield lour or live plates water). These rough-cut plates are the cost of the array We realized that, measuring 1 • 27 • 27 cm. The smaller crys­ "edged" on the diamond turning with this array design, the Nova tal is the size ol the typical boule produced machine. They are then mounted, 2(o/3(0 arrays would require more before large-aperture frequency-conversion than 500 individual KDP crystal technology using KDP was required. The stress free, with an elastomer onto a insen photograph shows finished KDP crystal rubber-gel chuck, and a flat reference plates, and fabrication would take plates (5 •• S cm, 15 ' 15 cm, and surface is machined. Using the crystal more time than the schedule 27 < 27 cm in dimension). orientation measurement system permitted. Therefore, we scaled up the (COMS), the angular correction is determined, and the crystal is then remachined to produce the final orientation and finish (Fig. 3). One of the 27 x 27-cm Nca KDP crystals being machined on the DL-1 diamond lathe is shown in Fig. 4. rystal Arrays During construction of the C Nova laser, single KDP crystals large enough to convert the 74-cm-diameter beam were not available, and it was not feasible to scale up the crystal technology in time to meet the Nova deadlines. In fact, the largest high-quality KDP crystals available at that time measured only 15 x 15 cm. Therefore, we were forced to design an array of KDP crystals. Our first array, tested on the Novette laser system, was a 5 x 5 matrix of 25 individual KDP crystals, each precisely machined so that its plate surface normal was within 30/irad of the phase-matched direction. However, because of the accuracy and careful handling required, this machining

Standard ^ /& EH* Boule Plates Edges Rough w grown cut finished crystal Phase match 2 1 obtained Fig. 3 (0** stags* of fabrication ol KDP crystals. Ws first I'JI the KDP boule ment system (COMS), the angular correction required for optimum con­ into plates. The plates are edga-fimshed and mounted on a rubber-gel version efficiency is determined. Finally (not shown), the KDP plates are chuck so that on* side can be machined flat Then the plate is reversed put back on the diamond-turning machine to correct the surface orienta­ and the other aide Js machined. Using tha crystal orientation measure­ tion and to mschlne the surface to its final optical finish.

47 size of each KDP crystal plate to Building an array of KDP crystals 27 x 27 cm, since crystals of this size for frequency conversion is difficult could be grown in time. This allowed because of the fine angular alignment construction of a 3 x 3 KDP array /or required between the Nova 1ft) pump the 74-cm aperture. This design was beam direction and the KDP less expensive and could be fabricated crystallographic axis. In previous in time to meet the 1984 Nova work using a single KDP crystal, the completion deadline. conversion efficiency would be maximized by manually adjusting the crystal alignment to reach the Fig. 4 optimum orientation. With an array of A KDP crystal plate (1 < 27 - 27 cm) im­ crystals, each element of the array mersed in an oil shower (for temperature sta­ must be aligned to the beam bility} as rt is machined on the DL-1 diamond individually. An approach using lathe. We use single-point diamond turning to individual crystal adjustments, orient the surface normal relative to the inter­ nal crystal structure and to machine the final although feasible, would be very optical finish. complex and is of questionable practicality on a large, complex system like Nova To avoid adjustments of individual crystals and align only the overall array, each of the crystals in the Nova arrays is precisely mounted so that when the Nova beam passes through the assembly, each crystal is aligned to within <100prad of the optimum orientation. It was fairly straightforward to bolt the precisely fabricated KDP plates to a support structure without complex alignment adjustments. However, this "egg crate" support structure, itself, had to be machined extremely flat so that when mounted, all the KDP crystals would lie in a plane to within better than 30/jrad. The Quadrature/ Cascade design required that both sides of the egg CTate be flat to this tolerance, one side for the doubler array and the other side for the mixer array. One very important feature of the egg-crate design is the extremely thin (3-mm) webs that provide the crystal supports. The webs are thin so they do not obscure a large fraction of the beam aperture. To provide the required stiffness and yet maintain the alignment tolerances, the webs are deep (100 mm). The egg crate is fabricated from a single piece of aluminum using electric-discharge Fig. 5 machining (HDM) technology to prevent warping of the thin webs. Ten A 3 ' 3 deep-web "egg crate," mounted on the OL-3 diamond lathe. As of the-Nova egg crates were fabricated the egg crste spins on the lathe, the single-point diamond tool traverses it, producing an extremely flat <<10 /an) surface. Both sides of the Nova on the DL-3 lathe, one for each beam egg-crate must be finished in this manner. line (see Fig. 5) In addition to the egg

48 INERTIAL FUSION

crates, special crystal installation coating design and a high damage fixtures were made to assist with the threshold." It can be easily applied loading of the 18 individual KDP by a spinning technique similar lo crystals. The final assembly was the photoresist deposition technique performed in a class-100 clean room, used in semiconductor technology.i: after which the crystal array was Figure 6 shows the coating tank used tested interferometrically for to coat each of the Nova KDP crystals. Rg. 6 wavefront distortion and crystal We optimized the spectral The sol-gel antiretlection coating is appiied to alignment, before it was installed on characteristics of this sol-gel coating the Nova KDP cryslals in the crystal-coating the Nova system. for each position in the array and tank shown here. The crystal is mounted on a coated all KDP crystals before they plate in the center of the tank. The crystal is A ntirefiection Coating were installed in the egg crate. The spun at about 3S0 rpm while the coating solu­ tion (0.75% 5i02 in ethanol) is applied from a £^L The final step in the final assembly is shown in Fig. 7. syringe mounted above the crystal. A ^L, development of the Nova frequency-conversion array was the invention of an antireflection coating for KDP. Uncoated KDP crystals mounted in a 2w/3w architecture lose approximately 16% in performance as a result of Fresnel reflections at each crystal surface. In previous frequency- conversion systems, this loss was reduced by immersing the crystals in an index-matching fluid. In fact, the first 2(0 array built for the Novette laser was a fluid-filled array. However, this method proved to be less than satisfactory.10 The fluid-filled gaps must be very narrow—a few microns—and the width is difficult to control. Also, the fluid-filled cells are subject to damage initiated by the interaction of the high-intensity light with microscopic contaminants in the fluid. The light heats the particles and induces local decompositions of the fluid. Slight opacities then develop, which absorb energy on successive shots and grow progressively larger, eventually obscuring a large fraction of the aperture. During the initial Nova design, as Fig. 7 a backup design to the low-loss but damage-prone fluid-filled cells, we An assembled, 74- cm-aperture. KDP built a bare crystal, deep-web egg- crystal array. Indi­ crate design and successfully tested it vidual KDP crystal* in a 4

49 erformance We identified the problem as an The crystal arrays performed intensity-dependent depolarization Pwell at low intensity during of the la) pump beam due to a tests on the Nova laser. However, combination of stress-induced when we increased the beam depolarization in the optical intensity, the frequency-conversion components and a nonlinear efficiency unexpectedly decreased. propagation phenomenon, called Above l.5GW/cm\ the conversion polarization ellipse rotation, that efficiency from i to to 3ra departed appears at high intensity. To solve this radically from our theoretical problem, we added a polarizer to the estimates and from the results of end of a Nova laser chain, before the small-scale experiments conducted frequency-conversion array, to prevent prior to the Nova/Novette trials. The the "contamination" of the beam by 2m performance also fell off at the weakly depolarized light. This same pump intensity. Closer increased the conversion efficiency examination of the early Novette 2m, significantly (Fig. 8). Residual 3d), and 4o? experiments revealed that, alignment errors introduced by the in all cases, the measured conversion prototype 74-cm-aperrure polarizer efficiency departed from our are believed to be the source of the predictions when the pump intensities slightly low conversion efficiency and exceeded — !.5 GW/cirr. can be eliminated by improving the Fig. 8 In small-scale, well-characterized optical quality of the polarizer Third-harmonic conversion efficiency vs input experiments conducted with Nova substrate. beam intensity observed on Nova beam 10, production KDP crystals, 2w and 35kJ, 1 ns). All ten Nova beams together can produce >50 kj of ultraviolet light in 1 ns and even more energy at longer pulse widths. At this stage in the development of large-aperture frequency-conversion arrays, there is no fundamental limitation on the continued scaling of the aperture in order to increase the output energy of future laser svstems. U

Kev Words: v££ i:jU' JITJ\: tri'qurniv ionvtTsu'fn. h.irmiinx urncrjliim—liu. 2to. 3fu. Ij'.fr— \in.) Vivt'tre. pnussium itihvLiri>m'n Intensity at 1m GW/cm7 phuKph.ili' {KI>PI. jviTisu>n maihininj;

50 No»« and References High Efficiency Third Harmonic Conversion of High Tower NdiGlass Laser 1. D, Eimerl, "Laser Fusion with Green and Radiation," Optics Communications, 34 (3), Blue Light." Energy and Technology 469 (1980). Review, Rept. UCRL-52000-82-8 (August 8. G. J. Linford et at., "Large Aperture 1982), p. 19-29. Harmonic Conversion Experiments at 2. P. A. Franken et al.. Phys. Rev. Lelt. 7. 118 Lawrence Livermore National Laboratory," (1961). Appl. Opt. 21 (2D). 3633 (1982): see also G. 3. J. A. Giordmaine, "Mixing of Light Beams J. Linford etai. "Large Aperture Harmonic in Crystals." Phys. Ret: Lelt. S, 19 (1962); Conversion Experiments at Lawrence see also P. D Maker et al. "Effects of Livermore National Laboratory; Dispersion and Focusing on the Production Comments," Appl. Opt. 12 (13), 1957 of Optical Harmonics." Phys. Rev. Lett 8, (1983). 21 (1962). 9. J. F. Holzrichter er al., "Research with 4. \. Nuckolls el at., "Laser Compression of High-Power Short-Wavelength Lasers," Matter to Super-High Densities: Science 229 (4718), 1D45 (1985). Thermonuclear (CTR) Applications." 10. M. A. Summers et at., "Nova Frequency Nature 239. 139 (1972) and Focusing System," Laser Program 5. 1977 Annual Report on Laser Fusion Annual Report, Lawrence Livermore Research, KMS Fusion, Inc.. Ann Arbor, National Laboratory, Rept. UCRL-50021-83 Michigan, Rept. KMSF-U762. pp. 2-17 (1983). p. 2-8. (1977). 11. D. Milam and I. M. Thomas, "Sol-Gel 6. F Amiranoff et al., "Interaction Coatings," Laser Program Annual Report. Experiments at Various Laser Lawrence Livermore National Laboratory, Wavelengths." presented at the Twenty- Rept. UCRL-5O021-B4 (1984). p. 6-39; see First Annual APS Meeting Division of also I. M. Thomas. "Optical Coatings by Plasma Physics, Boston, Massachusetts, the Sol-Gel Process," Energy and November 12-16, 1979. Technology Review. UCRL-5Z000-85-I0 7 R. S. Craxton. Theory of High Efficiency (October 1985), p. 8-14. Third Harmonic Generation of High 12. "Frequency Conversion for High Power Power Nd:Glass Laser Radiation." Optics Lasers," Technology Transfer Symposium, Communications. 34 (3). 474 (1980); see Lawrence Livermore National Laboratory also W Seka er al.. "Demonstration of (June 13-14, 1985). Eliminating Platinum Inclusions in Laser Glass

Damaging platinum inclusions in Nova laser glass can be eliminated by new glass- production techniques. The improved glass will replace the glass disks now in the laser amplifiers and will permit long-term, damage-free operation at full power.

For further information contact he large, neodymium-doped being developed. Replacement is Jack Campbell <«5> 422-6497. phosphate-glass disks in the estimated to take about 18 months. TNova laser amplifiers have Fortunately, since we have about been damaged because of microscopic, 24 months of experiments that require metallic-platinum inclusions. These only 50% power, we will be able to inclusions come from the platinum use Nova fully while the glass is crucibles used to melt the glass in the being replaced. production process. The inclusions Damage sites caused by the laser have a median size of only about irradiation of platinum inclusions have \0ftm, roughly equivalent to the size been observed in the neodymium- of a white blood cell. Although doped phosphate-glass components extremely small, they vaporize upon of both the Nova laser and its exposure to the powerful pulses of predecessor, Novette. The glass disks laser light and produce macroscopic cor .->in platinum particles with sizes fracture sites within the glass (see ranging from less than 5 n m (the Fig. 1). These fracture sites vary in size detection limit) up to about 100/im, from a few hundred micrometres tt; with a mean of about ID to 20 pm, several millimetres. and varying in concentration from Damage to the laser glass has been roughly 15 per litre to as many as found to occur at laser-light ftuences 1000 per litre. significantly below the Nova design The inclusions cause fractures in limit. Much of the glass will have to the glass when they are irradiated by be replaced if Nova is to be operated laser pulses of 1-ns duration above a without damage at its design limit. To threshold value of about 2.5 to prevent further damage to the laser 3.0J/cm2 We have found, on the glass, we have chosen to limit the basis of experiments and computer output of the laser to about 50% of modeling, that this damage threshold the design capability (even though corresponds to vaporization of the Nova has already demonstrated it can platinum particles as they are heated meet its full specified performance). to temperatures above the normal We will stay at this level until the boiling point of platinum. Results amplifier disks can be replaced with from both models and experiments the inclusion-free glass that is now show that the damage threshold INERTIAL FUSION

should be independent of particle size Nova, nor did they predict the growth (for inclusions larger than 1.0//m) and of the damage around the inclusions should increase as approximately the with repeated i- radiations. square root of the laser pulse length. One of the most disturbing xtent of Nova Class observations associated with the Damage platinum damage is that it continues E The output stage of a beam to grow with repeated irradiations. line of the Nova laser contains groups Therefore, for all practical purposes, of neodymium-doped glass amplifiers Fig. 1 all except very small platinum with apertures of 20.8, 31.5 and 46 cm A typical neodymium-doped, phosphate- inclusions (£1 pm) must be eliminated (see Fig. 2). Each group of amplifiers is glass, half elliptical disk from a large 46cm from the glass before it can be ased in followed by a vacuum spatial filter. A amplifier in Nova contains about 7 litres of glass; two half-disks are joined together to a high-power laser system. laser pulse passing along the beam form a full ellipse, (a) The damage spots in the line is amplified at one beam diameter glass are produced by microscopic inclusions istorical Perspective to the desired pulse energy and then of platinum. During our inspection of the disks Platinum-inclusion damage focused into a spatial filter that in Nova, we found between 100 and 10 000 damage sites in each disk. (b)One of the in laser and optical glasses removes small-scale spatial ripple and H enlarges the beam diameter to match 46-cm amplifiers, shuwing the location ol the has been a recurrent problem glass disks; the disks are set at Brewster's throughout the development of high- the next group of amplifiers. A spatial angle within the amplifier box. power lasers. In particular, platinum damage in silicate laser glass has been well known for many years. In some laser systems, platinum inclusions were not a problem. Our previous high-power laser, 5hiva, used a neodymium-doped silicate glass, and we observed no damage in that system. However, results from current studies of the Nova glass indicate that Shiva operated just below the damage threshold for platinum. We chose phosphate-based glass for use in Nova for a number of physics performance reasons, but we also expected it to reduce the risk of problems with platinum inclusions. Our expectation was based on the fact that the solubility of platinum is much greater in phosphate glasses than in silicates. When phosphate glass was selected for use in Nova, we tested five 15 x 15 x 2-cm samples with 1-ns, 1.064-pm pulses at 12J/cm2, using a 6-cnvdiameter beam produced by an arm of the Argus laser. These experiments indicated that the glass typically might contain one small (<10-pm) inclusion in each 10 cm3. Our estimate of inclusion density was correct for all but a few nf Nova's disks. However, these qualification studies did not correctly predict the large sizes (some exceeding 250 pm) of some of the inclusions that have been found in glass supplied for

53 filter at the output of the 46-cm probably replace the laser glass in the amplifiers further deans the beam and last 20.8-cm amplifier. increases its diameter to 74 cm, the diameter maintained over the final olving the Inclusion path to the target chamber. Problem As a result of this series of S In May 1985, a group of LLNL amplifications and beam expansions, chemists, material scientists, physicists, the laser fluence (i.e., the energy per and engineers were assembled to unit area) varies with position along address the problem of eliminating the chain (see the sawtooth curves in platinum inclusions in phosphate laser Fig. 2), with the largest fluence glass. The group had five major goals: occurring at the output of the final • To assess the damage to Nova 46-cm amplifier. The number of glass and to determine an operating range disks that must be replaced in Nova for which no further damage would can be determined by comparing the occur while improved glass was being measured damage thresholds with the developed. average fluences at various positions • To measure the damage along the beam line. Such a thresholds of selected platinum comparison is made in Fig. 2, in which inclusions and to determine the the shaded bands indicate the damage growth rate of the damage around thresholds for 1- and 5-ns-long laser the inclusions. pulses, It is clear that all of the disks • To develop a quantitative model in the 46-cm amplifiers and most of for the damage caused by laser those in the 31.5-cm amplifiers will heating of the platinum inclusions. need to be replaced with the • To develop a reliable quality- inclusion-free glass. As an added assurance technique for inspecting margin of safety, we will also large volumes of glass for microscopic (a)

Rotator Spatial 31.5'Cm amplifiers fitter HIu300l\300E '—' 4 5 6 7 8

Rotator Input to 20.6-cm amplifiers

16

Fig. 2 Saam fluance at normal incidence ae a function of position along th* law beam path, (a) Diagram ot ma bairn path through the final threa amplifier etagei along on* of the Nova bairn lines, (b) Laaar riuence for two levels of beam power. The horizontal band* indicate the threshold for glaaa damage from piebnum inclusions at 1-m and S-ns put** length*. (The diapfayed threahoMa are larger by a factor of 1.S2 to account for the fad that the gfaea disks are poartionad at Brewster's 20.8 cm 46.0 cm angle). The numbers on the absissa refer to the amplifier loca­ 31.5 cm tions shown in th* beam-path diagram (a). Amplifier number

54 INF.RTIAL platinum inclusions (see the article on Although the microscopic p. 24). examination distinguishes among • To work with the glass suppliers damage sites, bubbles, and to quantify the mechanisms for undamaged inclusions, it cannot platinum contamination and to identify the chemical composition of develop manufacturing processes to the inclusion. To verify that the eliminate or minimize it. inclusions were indeed platinum, we All five of these tasks have been cored cylindrical samples of several of completed, and we are in the process the damage sites and then fractured of procuring the replacement laser the core through the inclusion area. glass for Nova. Using scanning electron microscopy and electron-probe microanalysis, we A ssessmenr. of Glass found a high concentration of ZA Damage platinum in the damage sites. In some X jL All of the beam lines on cases, we also found a pure platinum Nova have had tens of shots that kernel remaining from the initial produced energies in some sections of inclusion. the laser greater than the thresholds Figure 4 displays the results from for damage by platinum inclusions. examination of several of the 46-cm Consequently, one of our first tasks was to assess the damage level in Direction of illumination the present Nova system and to Fig. 3 determine the operating power for Schematic representation ol the morphology which no further damage would occur of glass damage caused by a platinum inclu­ sion. There is damage in a small tobe in either with the existing ^lass. direction along the beam, but it is not as To assess the extent of the damage extensive as the radial pattern. in the existing amplifier disks, we began a microscopic examination of selected disks from the various beam lines. We minimized interference with normal laser operations by examining a small but statistically representative number of all Nova glass disks. These disks were removed from the beam line and taken to an inspection station in a clean room. This inspection stand permits both back and side illumination of the disk with a focused beam from a high-intensity lamp. Large damage sites or inclusions are easily located by the light they scatter. Each damage site was then marked and examined with a microscope to determine its size. Using this method, a trained observer can locate most platinum inclusions or damage sites greater than 10 pm. Because of the characteristic fracture 200 400 600 BOO patterns about the damage sites, we Size of damage site (maximum dimension), jjm can easily distinguish among damage sites, bubbles, and undamaged Fig. 4 platinum inclusions. Most damage Normalized plot of the fraction of damage sites (in the 46-cm disks) that sites are distinguished by a series of exceed the dimension shown on the abscissa. The amplifier numbers radial fractures that form a circular correspond to the locutions shown in Fig. 2a, and the average fluence is pattern perpendicular to the beam given in parentheses. Many of the damage sites are greater than 500 um direction (see Fig. 3). in diameter In the amplifier disks exposed to lluences greater than the 3 5-J/cm3 (1-ni) damage threshold, i.e., ampifier disks 10.11, and 12. laser disk',. The data have been inspected the inclusion with an optical normalized and plotted to show the microscope. If there was no apparent fraction of damage sites greater than a alteration in the inclusion, we particular size. (We did this because irradiated it again at a higher fluence. the actual number of sites in a Once damage began, the fluence in nominal 7-litre laser disk varied fiom 1.3-ns experiments was increased to about 100 to as many as 10 000.) As 5 to IOJ/OTT, a range typical of the the data show, there is a large spread highest fluences anticipated in Nova. in the sizes of the damage sites. The inclusion was repeatedly Furthermore, we observed that the irradiated at these fluences until the median damage size increases nearly size of the damaged volume either linearly with increasing laser fluence. stabilized or reached a dimension We believe this linear relationship larger than 250pm. In experiments may be due only to the fact that the with either 9- or 50-ns pulses, the laser disks have seen a limited fluences used to observe evolution of number of shots. It is not clear what damage were selected to be 2 to 2.5 the functional relationship will be times the threshold for initiation of once the damage growth has damage. stabilized. The damage threshold was deiined These data provide an assessment as the appearance of one or more only of damage that has occurred to fractures extending a few micrometres the laser amplifier disks over the into the glass surrounding the course of a relatively small number of inclusion. These initial fractures shots, i.e., about 100 shots vs the had shapes that were peculiar to expected 10 000 shots over the life of individual inclusions. After a few Nova. However, it appears that all irradiations at fluences above amplifier disks that see fluences (at threshold, the damage usually evolved Brewster's angle) greater than into a circular, planar fracture oriented 3.5J/cm' (1 ns) will experience an perpendicular to the incident beam. unacceptable level of damage and This damage morphology is the same must be replaced. as that observed on Nova. We could not determine by optical microscopy easurements of whether the damage expanded along Damage Initiation the direction of the beam. M and Growth In most of the experiments, We measured threshold fluences for repeated irradiation of the inclusion fracture initiation in glass containing caused damage growth to sizes larger platinum inclusions and repeatedly than 250 pm. Our data for the largest irradiated the inclusions to determine inclusion we studied (major dimension the rate of damage growth. 75 pm) illustrate this damage growth. Undamaged glass samples containing As Fig. 5 shows, the inclusion platinum inclusions were cut from a survived three irradiations at fluences 20.8-cm phosphate-glass amplifier up to 2.4)/cnr and was first damaged disk. The major dimensions of the at 3J/crcr, during the fourth shot. inclusions ranged from 4 to 75 pm. Damage grew rapidly during the next Selected samples with inclusions were two shots at higher fluence but ceased placed in the center of a laser beam when the fluence was dropped below and irradiated at normal incidence 2.5 J/cnr. Four additional irradiations with 1.064-pm pulses having at higher fluences caused further durations of 1.3,

laser pulse. The characteristic length the model, laser light is absorbed on for heat flow into a platinum particle the upper surface of the platinum is only 0.15 fitr\ in 1 ns and 1.1 fim in sphere with a cosine distribution to 50 ns. Since all inclusions tested had account for the curvature of the dimensions exceeding 4 j/m, only the surface. illuminated surfaces of the particles were rapidly heated to the (a) vaporization temperature 1000 Damage growth • hermal Modeling • ol of Damage • TOur experiments and • • • calculations shoiv that laser light is • absorbed by the opaque platinum, — a causing a thin skin of platinum on i • • • the front surface of the inclusion to vaporize. This produces a shock wave II" that propagates through the inclusion and is transmitted to the glass. Glass, I I I I I I being a brittle material, fractures because of the stresses caused by the shock wave and the vaporized platinum. These fractures propagate until the pressure of the vaporized platinum decreases below that required for further crack propagation. Platinum damage in laser glass was first modeled in 1970 by Hopper and Uhlmann,' who used analytic expressions for the heat transport and 1 2 3 4 5 6 7 8 9 10 11 12 calculated temperatures exceeding Shot sequence 10 000 K when fluence levels were 20 J/ciTr over a 30-ns laser period. (b) They theorized that very small Initial appearance After shot 4 After shot 6 inclusions were safe at these pulse lengths because heat losses to the glass kept the temperatures and hence the stresses within allowable bounds. However, they argued that for sizes above 0.1 /m\, stresses exceeded the strength of the glass, and it could After shot 10 fracture. Their thermal modeling suggested that damage would be dependent upon particle size because the surface-to-volume ratio of the inclusion decreased as the radius ft increased. Therefore, the larger inclusions had more thermal capacity / / and hence lower average temperatures and stresses. Fig. 5 In our thermal model, we represent Evolution of damage surrounding a bar-shaped inclusion with initial ma­ the physical system with a sphere of jor dimensions ol 75 /ffn (one of the largesi we observed), (a) Graphs platinum surrounded by an annulus showing damage growth and laser fluence over the l2-$hot sequence. Damage began during shot 4 at 3.8 J/ems (normal incidence) and grew of glass (Fig. 6a). The model represents 1 rapidly during each shot at lluences greater man 4 J/tm . what could be a typical inclusion (b) Photographs of the area surrounding the inclusion clearly show the embedded in laser glass (Pig. 6b). In growth of the damage.

57 Literature values for the spectral platinum with negligible contact Fig. 6 and thermal properties of platinum resistance. The equation of state for Thermal model used to calculate the elfeci of and glass are used in the model. The platinum above the boiling point is laser heating on a platinum inclusion in phos­ surface of the platinum is allowed to that used in the LASNEX code. phate glass, (a) Conceptual diagram of what radiate to a 300-K (room temperature) We performed the modeling on a could be a typical inciusion and (b) our spheri­ surrounding medium. No absorption cal model representative of this inclusion. The Cray computer, using TOPAZ, a calculational zones are finest at the site of is permitted in the glass, but heat is finite-element, heat-conduction code, greatest transient response. conducted between the glass and the along with (he MAZE and ORION codes to prepare the input and process the output. The results of our (at (b) 1-/m laser light modeling are shown in Fig. 7. where the damage threshold is plotted w M u n i against laser pulse length. Figure 7 also shows our experimental Phosphate glass measurements (black dots) of damage Platinum thresholds at the three pulse lengths, 1.3. 9, and 50 ns; the damage thresholds were 2.5 ± 0.3, 4 x 0.4, and 8 ± 0.8 J/cnr. respectively. Phosphate The solid line that represents our glass model in Fig. 7 has a slope of about 1/2, which corresponds to approximately a square-root !oruni zoneo —^; dependence of damage threshold Fine zoned-j^ If on laser pulse length. The slight Mediu.Hii im zoneTAnffrdi -~ '. departure of the fheuiy from the data at the shortest pulse length (1.3 ns) mav reflect the limited crack propagation per shot in this region. In other words, it takes a number 100 , of shots before any damage can be - Data physlcallv observed. Results from • LLNL our model compare well with other 4. Avizpnjs and Farrington3 experimental data, which are also • Yamanaks el at? shown in Fig. 7.

4 • National Materials Advisory Board Figure 8 shows the thermal profiles that are predicted by our model for a platinum inclusion in glass irradiated 10 — .^i L with a 1-ns pulse of 1.06-4-^m laser I light at a fluence of 3.7J/cm:. The i thermal profiles are for two time periods, 1 ns (immediately at the end of the pulse) and 10 ns (9 ns after the

Model^-"^^ i end of the pulse). The profile at 10 ns shows the cooling that occurs at the inclusion surface, largely as a result of thermal conduction into the platinum. i i i i 1 i i , I There is negligible heat loss into the 10 100 surrounding glass over this time Laser pulse length, nj period. Peak temperatures at the end of the pulse are high (greater than Fig. 7 10 000 K), suggesting that significant Effect of pulse length on damage threshold fluences. The data points vaporization of platinum is occurring represent measurements made by LLNL researchers and by experiment­ at the inclusion surface (the normal ers elsewhere; the solid line represents the results of LLNL model cal­ culations. Agreement between our calculation and the experiments is boiling point of platinum is about quite good. 4100 K).

58 INERTIAL FUSION

Table 1 lists the calculated independent of inclusion size (above maximum platinum surface about 1 /jm) and the threshold for temperatures and resulting pressures damage initiation is lower for the of vaporized platinum for laser shots shorter laser pulses. at various fluences. We have no doubt that the pressures are high enough to lass-Melting propagate fractures. This, in Technology conjunction with our experimental G As part of our effort to results, suggests that damage will eliminate platinum inclusions in the occur if the temperature of the platinum exceeds its boiling point, The 10 experimental data in Fig. 7 show a Fig. 8 : (a) -Glass damage limit of about 2.5J/cm for a Calculated temperature contours for a 5-um- ]-ns pulse and 4J/cnr for a 10-ns radius platinum inclusion surrounded by phos­ pulse. This agrees fairly well with phate glass and illuminated by a 1-ns pulse of our calculated values. 1.064-/on laser light at an Incident lluence of 3.7 J/cnr*. (a) Calculation at 1 ns (i.e., immedi­ We are also developing a mode! for ately after the pulse) showing the strong sur­ the glass fracturing induced by the face heating of the particle; the temperature heating of the platinum particle. We contours range from a minimum of 1500 K to a have not yet developed it sufficiently maiimumol 11 000 K. (b) Calculation at 10 ns (9 ns after the pulse); the temperatures have to predict crack growth. However, i»h cooled considerably (minimum of about 570 K, crack growth can be conservatively maximum of 2700 K) and the heat has pene­ estimated by multiplying the time trated 1 to 2 um into the particle. during which platinum is in a vaporized state by the speed of sound 15 10

Table 1 Calculi ted characteristics of damage to gliss caused by metallic platinum inclusions. The predicted fluence damage threshold is 1.4 J/cm1 for a 1-ns User pulse length.

Crack Incident Absorbed Pulse Maximum Maximum propagation fluence, fluence, length, temperature. pressure. per shot. J/rm' l/cm1 ns K GPa um

37 in l 3 i in' 280 500 16 5 76i 111* 150 15H 3.7 l 1 5 x 111' 50 H-] 1.8 0 5 h.5 i 10' -50 35 1.5 114 5 3 x 10' •'50 l 5 3.7 1 10 15i mJ — -

59 phosphate glass used in the Nova contamination generally does not laser, we are working with the two affect optical performance. companies that are supplying the glass The sources of platinum-particle (Hoya Optics, inc., and Schott Glass contamination in glass have been the Technologies, Inc.). Our efforts are subject of several previous studies/''' It directed toward understanding the is generally agreed that the particles formation mechanism of platinum are produced by abrasion or wear, by inclusions in phosphate glass and direct dissolution of the platinum developing process conditions to crucible and subsequent precipitation eliminate it. of the dissolved platinum, or by vapor-phase transport via a volatile

Preparing the Laser Glass oxide (PtO;) and redeposition as The laser glass currently in use in metallic platinum. In principle, Nova was prepared in a two-step, abrasion and wear can be eliminated batch-melting process. In the first step by care in processing and have of this process, a cullet glass is made usually been ruled out as a source by melting the raw starting materials. of contamination. This cullet then serves as the Platinum is an unusual metal in feedstock for the final melting of the that it forms a volatile oxide (PtO,) optical-quality glass. The cullet is when heated above about 700°C in an usually melted in a relatively inert, oxygen-containing atmosphere.'" The refractory crucible. The molten cullet PtO; vapor pressure in equilibrium

is poured onto a cold surface wheTe with metallic platinum (at 0.1 MPa 02) it fractures into smaller pieces as a is about lO3 times greater than that of result of the thermal stresses. The the platinum metal. At typical melting characteristics of the cullet are far temperatures, PtO; is stable only in different from those of the final gas phase: upon contact with a solid product; the cullet is generally full or liquid surface, it redeposits of bubbles and striae and possibly platinum metal.'1 contains small particles of undissolved Therefore, ir. oxidizing systems, it is component material. possible to transport platinum to the

The fractured glass that is produced molten glass via PtO; and then from the melted cullet is used in the redeposit it as platinum metal. This second and final melting. This step is mechanism is further confirmed by carried out in a platinum crucible of experiments in which the glass is about 4D-!itre capacity. The use of melted in a ceramic crucible that is platinum vessels has long been the physically separated (but in close standard in the glass industry because proximity) to platinum metal.3 When platinum is erosion resistant and the melt is carried out in oxygen, essentially insoluble in the glass platinum particles are found in the matrix. During the final step, the glass. On the other hand, when an molten glass is mechanically stirred to inert gas is used (e.g., nitrogen), no homogenize the glass and to remove platinum particles are observed. It any striae produced by volatilization. would appear, therefore, that one It is also treated to remove bubbles. could eliminate platinum-particle At the end of the final melt, the glass formation by preparing glass in an is cast into a mold and annealed. inert environment. Unfortunately, the oxygen released by the decomposition Chemistry of Platinum of oxide components of the glass* may Even though platinum is largely be sufficient to produce platinum inert to most molten glasses, it still transport and subsequent inclusion causes a low level of particulate formation. "* contamination, generally at a level A related study" of silicate laser of less than a few parts per million. glass reports that platinum solubility Except for the glass used in high- increases with oxygen content of the power lasers, this platinum gas during the mell This study INERTIAL FUSION

proposes that platinum dissolution (by riarinum dissolution appears to be the glass) and later precipitation may a means of eliminating platinum be the mechanism causing inclusion inclusions, but there are several formation in melts carried out in inert limitations to this method. First, of or reducing environments. course, is the solubility limit of Unlike silicate glasses, phosphate platinum in the glass. Fortunately, for glasses are known to have a much phosphate glasses, the solubility limit higher solubility limit for platinum is quite high (>1000ppm) under (several orders of magnitude higher); normal processing conditions. that is, phosphate glasses can hold Therefore, the glass continues to hold much more platinum in solution the ionic platinum even while cooling before it precipitates as metal particles. down, and precipitation at that point It was generally thought, therefore, generally has not been a problem. that the platinum would remain in A greater concern than the solution in the glass and thus that solubility limit is the absorption of platinum-inclusion formation would 400-nm-wavelength light by the ionic not be a problem in phosphate platinum dissolved in the giass. glasses. Absorption of light of this wavelength reduces the pumping efficiency of the Key Process Variables xenon flashlamps and could affect the A number of variables affect the laser's performance. We have found formation of metallic-platinum that as long as the absorbance at 400- inclusions in phosphate glasses. The nm remains below 0.2 cm ', laser most important are the oxygen performance is not affected. This content of the process gas. the melt absorption level corresponds to an temperature and temperature ionic-platinum concentration of uniformity, the processing time, and about 130 ppm. the platinum solubility in the glass. Schott has found that the Experiments were undertaken at both processing temperature also Hoya Optics, Inc., and Schott Glass Technologies, Inc., to examine the effects of these variables on the density of platinum inclusions in their 1 r- glasses. Both the Hoya and Schott scientists observed a dramatic reduction of inclusions in the phosphate glass when the oxygen HJO^ mixture content of the processing gas was increased. This result is illustrated in Fig. 9, in which platinum-inclusion count is plotted as a function of oxygen partial pressure in the process gas. These data are from 0.5-litre test melts of glass carried out at Hoya; a -3 li i i i 10' 10 1 10- 10' similar trend was observed at Schott Oxygen, % for their particular glass composition. Hoya also found that use of a Fig. 9 proprietary additive further reduced the inclusion count under the most Relation between oxygen partial pressure in oxidizing conditions (also shown in processing gases and measured density of platinum inclusions. The results (data Fig. 9). The experiments at Schott and points) are for half-Mr* melts carried out at Hoya clearly demonstrate that the about 1100°C. The black line shows (he re­ oxidizing conditions promote the sults for an oxygen nitrogen mixture. The dissolution of platinum particles upper line show* the effect of adding a pro­ in the melt. prietary additive developed by Hoya Optics, Inc., to the process ga»

61 dramatically affects the inclusion 400 nm is much less dramatic than the density in their glass (see Fig. 10). effect of temperature. The fact that Raising the processing temperature time has less effect than temperature from 1100 to H50°C produces about a is not surprising; the rate of ID5 to 10* drop in inclusion density. dissolution is governed by a typical These results are for a series of half- Arrhenius relationship that depends litre melts in the presence of oxidizing linearly on time and exponentially process gas. For this same series of on -1/7; where T is temperature. melts, Schott also measured the Changes in process temperature, absorption at 400 nm and found therefore, produce a much larger that it remains below the 0.2-cm ' effect than corresponding changes in specification. processing time. The effect of processing time on In parallel with the dissolution inclusion reduction or absorption at experiments by the glass vendors, we have also developed a simple one- dimensional model of the process Fig. 10 by which the platinum inclusions Measured inclusion nanaity aa a function of dissolve in the glass. Using this tfw temperature al which phosphate glaas ia model, we have calculated that the melted (the eiperimental temperature range dissolution time for 10- to 20-fim was 1100 to 1450°C). These experiments were carried out by Schott Glaaa Technol­ metallic platinum particles is about ogies Inc. 2 to 10 hours at 1100°C (Fig. 11). This calculated result agrees well with experiments by the glass companies. One of the conclusions from our calculations is that the dissolution time is proportional to the square of the particle size. Therefore, trying to dissolve very large particles (lOC/im diameter) is not possible * -ith the current processing method. Fortunately, inclusions greater than 20 fim diameter are rare and should 1100 1200 1300 1400 not present a problem. Temperature,°C Using results from these and other

s experiments, both Hoya and Schott

10 1 Fig. 11 have been able to modify their Calculated dissolution limes for platinum in­ melting processes to eliminate most if clusions in glass as a function ol particle size. not all platinum inclusions. In brief, Most inclusions tie about 10 to 20 )m in diam­ both companies have sought to eter and are estimated to lake about 2 to control the platinum problem by 10 hours lo dissolve. We obtained these re­ / Range of sults with a simple, one-dimensional model 10* - eliminating outside sources of we developed to simulate platinum particle / size lor platinum contamination and by dissolution. - / most adjusting process conditions (e.g., / inclusions oxygen partial pressure, temperature, , and time) to place and keep all I platinum in solution. 5 10 rr / eplacement of Nova Laser Glass Considerable progress has • / R been made toward the production of inclusion-free glass. In early December l—L.l 1 1 L_ Ul 1985, .we tested seven large pieces of 10 10J glass with a total volume of about 17 Particle size. f*n to 18 litres (equivalent to about three

62 INERTIAL FUSION

large laser disks) and found an originally supplied. We estimate the average of about 0.2 inclusions per replacement will be complete litre of glass; four pieces of glass were approximately 18 months after the completely platinum free, for start of production melting. L3 comparison, the glass disks now in Nova have platinum-inclusion concentrations of between 10 and Key Words: computer code—f.AZE. ORION. 1000 per litre; thus the new glass TOPAZ: glass—laser, neodvmium. phosphale. represents about a 100-fold or greater platinum inclusion, siliate: laser—amplifier, improvement. This would yield large damage, glass. Nova laser disks with an average of only 1 to 2 particles each. (The 31.5-cm and 46.0-crn disks each contain about 7 litres of glass). Using both Nova and Notes and References our new rapid-scan inspection tool I. K IV Hopper and D R. Uhlmann. (see the article on p. 24), we tested the Mechanism of Inclusion Damage in Laser samples for damage at two to four Glass." I. Appl. Ptiys 41. 4O23-J037 (1970) times the fluence of the damage 2 P V Avisonis and T. Famngton. Internal threshold. Self-Damage of Ruoy and \d-Glass Users," Appl. fhys. Lett. 7 (1965). Hoya and Schott have both 3 C. Yamanaka et al.. //u'esrigarion of embarked on programs to construct Damage in User Class. National Bureau of new melters embodying the results of Standards, Special Publication 356 (1971). their experiments in platinum-particle •). National Academy of Science. National Academy of Engineering. Fundamentals of reduction and including enhancements Ojmjge m Laser G/as.s, publication which should further reduce \MAB-271. Washington, D.C. (1970). platinum-particle content—the goal, 5. R. I- Ginther, "The Contamination of Glass of course, being none. We anticipate pv Platinum." / Xonen-stalllne Solid* b, initiating, in mid-1986, pilot runs of 294(1971). about 20 large disks produced by each fv E I Homyak and R P. Ahendroth, rmbhrns of Melting C/ass in Platinum. vendor using the new melting Owens-llinois, Inc.. Toledo, Ohio (1970). facilities, At lhat time, we will request 7. |. C. Chaston, Platinum Metals. f?ev;eiv 19 a price quotation on production (•1). 135 (1975). quantities of the glass. Our plan is 8. C. B. Alcock and G VV. Hooper. to obtain from each company "Thermodynamics of the Gaseous Oxides of the Platinum-Group Metals." Proc. Roy replacement glass for the disks it Sor A 254. 551 (19ftC!|

63 Detecting Microscopic Inclusions in Optical Glass

Our automatic system for detecting potentially damaging inclusions in optical glass can find and identify microscopic particles in the parts-per-quadrillion range.

For further information contact asers of high peak power, inclusions that cause damage are in John Marion (41il 42.V67HH. such as those used in fusion the micron to submicron range—often L research and proposed for use too small to be reliably detected by in possible defensive laser weapons, conventional optical microscopy. Light require large glass optical components scattering techniques also are (10 litres or more of glass). These unwieldy for inclusions al such low- components must withstand laser densities {< 10 Vcm1) because intensities in excess of 1 GW/cm2. spurious signals from bubbles, surface Thus it is essential that they be free of imperfections, and other scattering inclusions, which can cause damage to sources swamp the signals of interest. the glass under high laser fluences We have developed a novel At present, most glass optical approach for optical inspection in components are melted and refined in which we use the focused output from platinum crucibles. These platinum a commercial, Q-switched, yttrium- crucibles are the source of the aluminum-garnet (Nd:YAG) laser to microscopic particles that can become scan the entire glass component for included in the glass. When exposed the presence of inclusions. The energy to high-energy laser irradiation, these of the scanning laser beam is platinum inclusions partially vaporize sufficient to partially vaporize the and cause cracks within the glass that absorbing inclusions, producing a interfere with the propagating laser plasma (- 10000 K) These tiny beam. inclusions are readily detected by the Our current research is directed visible emission from this plasma; a toward reducing or eliminating photomultiplier tube is used to resolve these inclusions in future optical the light emissions A computer components (see the article on p. 12). records these events and generates Reliable detection of such inclusions is an inclusion map of the optical a prerequisite to developing optical component. This scanning technique components thai are highly resistant reliably detects micron-sized to damage. Unfortunately, traditional inclusions in large volumes of glass detection methods have not been and greatlv reduces the long times adequately successful. Y-or the laser and human subjectivity inherent in energies of interest, the size of the previous visual inspection techniques.

64 INERTIAL FUSION

One previous method of testing length and enters a beam splitter optical glass for damage, used by whose front surface is coated to researchers elsewhere, was to focus a minimize reflection. The back surface small (3-mnr or smaller) laser beam of the beam splitter is imaged onto a Fig. 1 at high fluences on a few areas in a calorimeter whose output is used to Schematic drawing of the rapid-scan inspec­ 1 sample. The results often were continuously monitor laser tion system tor automatically detecting metal­ coupled with a statistical treatment of performance. lic inclusions in optical glass. the data to yield an overall measure of defects.2 Unfortunately, this method of inspection and damage assessment is completely ineffective for glass with compute* a sparse distribution of defects. In another approach to inspection of Printer Interface large optics, we used the Nova laser to give high fluences over a large 3 II -- 5m) area ; however, these tests were Turning restricted to several shots per day mirror Commercial Net YAO and required a subsequent optical 4*m inspection. With our new rapid-scan inspection device, we have successfully inspected 10-litre (>104-cmJ) optical components and have detected Calorimeter inclusions at levels in the parts-per- Commercial quadrillion (lO13) range. The $300 000 x-z scanning stage cost of the entire assembly is modest, . The profiles are reasonably Gaussian.

65 After passing the splitter, the laser the component to ten shots at more beam strikes the sample. The sample, than twice the damage threshold for mounted on a commercial stage drive platinum inclusions in glass.4 After capable of movement along both x the inclusions were irradiated, the and y axes, is scanned continually as resulting damage was sufficiently it is moved in front ol the laser beam. large that we could detect the sites The entire rapid-scan detection system by visual inspection using side is controlled by a laboratory computer illumination of the sample. This new and interfaced through a general- detector has greatly streamlined our purpose data-aquisition controller. inspection of optical glass. With our present detection system, the irradiated area of the sample is A dditional Applications imaged onto a photomultiplier tube, i^k Our new testing device is and the output is digitized and J. A. also being used by the monitored by the computer. Optical Laboratory's Laser program to study signals emitted by the plasma (from the effects of multiple laser shots on the heated inclusions) are combined various types of defects. With this with x- and y-position coordinates and rapid-scan system, we can readily used to generate an inclusion map accumulate data on the growth rate (Fig. 3). of damage from individual defects Before we developed our new for numbers of shots far exceeding detector system, we inspected for projected requirements. inclusions by exposing each point in The rapid-scan system may also find applications in chemistry and materials science for the detection of (a) Inclusion map Fig. 3 small concentrations of microscopic Example results of inspection for inclusions in . inclusions or defects. To do this, two 10 optical glass: (a) computer-generated inclu­ -ftp- requirements must be met. First, the sion map produced by the rapid-scan inspec­ host material must be substantially tion device, and (b) map showing findings in S the subsequent optical inspection. Sites transparent at a wavelength accessible marked "1" are platinum inclusions, sites by high-power lasers, and, second, the marked "B" are bubbles, and sites marked : !&• defect or inclusion of interest must "C" are chips at the edge of the sample. absorb light significantly at that

±u- •* [ wavelength. Our method could be used, for example, with a Raman- ft shifted argon laser to detect metallic inclusions in semiconductor devices. It could also be used with a Raman- , , 4 shifted Nd:YAG laser to detect silicon 0 2 4 E 8 10 or carbon inclusions in Si3Nj or SiC x distance, cm structural ceramics. b) Optical inspection ummary I We have devised a simple • S system to inspect large optical components for potentially damaging microscopic metallic inclusions. Our t 1 rapid-scan detection device, which can • e B B deled inclusions in the parts-per- * * quadrillion concentration range, uses a i Q-switched NdrYAG laser focused on * 1 the optical component, which is • continually scanned during the test. The energy from the N'd:YAG laser C C -BOB IS causes any metallic inclusions in the x ejection glass to partially vaporize; these

66 plasma emissions are detected by a Materials (National Bureau of Standards photomultiplier tube, whose output Special Publication 35&. 1971), p. 76. is monitored by a computer The 2 L C. de Shazer, B. E Nevvnam. and K. M Leung, "The Role of Coating Defects in computer uses the output to generate Laser-Induced Damage to Thin Films.' an inclusion map. With our new Laser-fnduccd Damage in Optical detection system, it is now possible Material* (National Bureau of Standard* to inspect large optical components Special Publication 421, 1973). p. 114. rapidly, reliably, automatically, and at 3. D. Milam, C VV Hatcher, and |. H. moderate cost. J Campbell, "Platinum Particles in the \d:Doped Disks of Phosphate Glass in ihe Nova Laser." Proceedings of the Boulder Key Words: damage—materials study, optical Damage Symposium. Boulder, Colorado. component: glass—inspection, laser, optical, October 1985 (National Bureau of laser—Nd:YAG: Nova, Q-switched; platinum Standards), in press. inclusion. 4. J. £. Marion. C. J. Creiner, P. H. Chaffee, and J, H. Campbell, A Versatile LaseT Glass Inspection and Damage Testing Notes and References Facility," Proceedings of the Boulder 1. M. Bass and H. H. Barrett, The Probability Damage Conference. Boulder. Colorado. and Dynamics of Damaging Optical October 1985 (National Bureau of Materials with Lasers," Damage in Laser Standards), in press. Auxiliary Target Chamber for Nova We have added a second target chamber to the Nova laser facility to increase its experimental capabilities and to allow more cost-effective use of the laser. We have effectively doubled the productivity of the Nova laser for a fraction of the cost of the original Nova project.

For more information contact he basic Nova laser facility, 200-^m-diameter spot onto the target Grcp Susld U1SI 412-0681. completed in December 1984, at the center of the chamber. T was constructed primarily to These major subsystems are perform experiments in weapons supported with other equipment to physics and inertial confinement maximize the effectiveness and fusion. In its initial configuration,1 simplify the operation of Nova. We Nova was composed of two use an array of laser-energy and fund-mental subsystems, the laser image sensors to measure energy and itself and the target area, which to assist in beam alignment at several includes the instrumented 4.6-m- points along the laser chains. Target diameter vacuum target chamber. The diagnostic instruments, in many cases laser subsystem comprises ten chains uniquely designed for use on Nova, of cascaded amplifiers that collectively precisely measure the effects on generate approximately 100 TW of targets of the intense laser light during infrared (1.05-pm) light in a 1-ns time intervals of less than a billionth pulse. Each of the ten chains produces of a second. A power-conditioning a 74-cm-diameter beam of infrared system includes a 60-MJ, 20-kV light at its output. Large-aperture, capacitor bank that energizes the laser computer-controlled mirrors steer amplifiers and other high-power, these beams toward the main target electo-optic devices. An integrated chamber, where they arrive in two computer-control system uses a opposed conical clusters of five beams network of more than 50 computers each. The infrared light is converted to to coordinate setup, operation, and green light (0.527 um) or ultraviolet diagnosis of the entire Nova system. light (0,351 /Jm) by precision arrays Planning and construction of the of KDP (potassium ciihydrogen Nova laser system took eight years phosphate) crystals placed in the and was completed in December 1984. beam paths just before the full- By the end of l-ebruary 1985, we had aperture focusing lenses. These lenses begun operation of Nova as the provide a vacuum-tight window into world's most powerful and best the chamber and focus the beam instrumented facility for studying the down to a precisely positioned interaction of high-intensity laser light

68 with targets. As Nova was nearing To make productive use of Nova's completion, we began to set our plans "excess" laser capacity, we added the for the highest priority experiments; it smaller Shiva target chamber (the soon became evident that the facility same one that had been used with was becoming overbooked by a Novette) to the Nova facility as an growing number of important target- auxiliary chamber. This target experiment campaigns. chamber has allowed us to perform Our previous laser facility, Novette. experiments in one chamber while the had been shut down and its parts other is being prepared for different installed to complete the Nova experiments. The operation of the system. Novette had been assembled second chamber is fully integrated in 1982, "borrowing" two of Nova's into the Nova system. As a result, we ten laser chains, to perform target have effectively doubled our target experiments in a 2-m-diameter experiment capacity and flexibility, chamber salvaged from the Shiva and we are able to operate this second laser (dismantled in 1981). During chamber with only a modest increase Novette's relatively short lifetime, it in operational resources over those proved the value of a smaller, lower required for single chamber operation. power (relative to Nova's ten beams) We chose a Novette-like, two- target facility for certain types of beam, target-illumination design for experiments. We knew that we could this target chamber. Given the success continue these important experiments of the soft-x-ray laser experiments using the full Nova system. However, during the final days of Novette some of these experiments did not operation, we elected to optimize the require the full power of all ten Nova initial configuration o! the second laser beams, and to perform them in Nova target facility to support a Nova's large ten-beam chamber continuation of these studies. These would take away valuable svstem soft-x-ray laser experiments require time from the experiments that did that the high-intensity laser light be require its full capability. An example focused in a thin line (not a circular of a series of such experiments was spot). Th s creates an x-ray lasing the soft-x-ray laser campaign medium ii. the target with a performed with Novette in the sufficiently long path so that summer of 1984. This campaign significant gain can be measured. succeeded in producing the first verifiable x-ray laser in a laboratory Another important choice was the environment. As Novette was being wavelength of light to be used for shut down, scientists were already target experiments. The frequenrv- planning further experiments to conversion arrays for the ten-beam expand their knowledge of soft-x-ray chamber were designed to maximize lasing and its future applications. delivery of third-harmonic ultraviolet light." However, the soft-x-ray laser To address the demand for more experiments were planned for second- experimental capacity, we investigated harmonic green light, so the the limitations on the effective target frequency-conversion arrays for the experiment rate for Nova and two-beam target chamber were determined that, in keeping with past optimized for green-light conversion, experience, target-related factors while still providing acceptable dominate the laser-related factors. conversion efficiency for ultraviolet We noted that target-experiment light. campaigns frequently require different calibrations and setups of diagnostic ystem Design Overview instrumentation and beam-focusing To minimize costs, the two- arrangements. The full Nova laser can beam target chamber was be readied for firing more quickly S designed to make maximum use of than the diagnostics and targets can existing components. Major elements be installed and aligned. available from previous laser systems. primarily Novette, were reused where and propagated in a similar fashion possible. These included the target into the chamber using two, full- chamber itself (previously salvaged aperture, motorized mirror elements. from Shiva), the supporting space The 50-m beam path from the frame, parts of the larget-chamber optical switchyard to the mirrors in vacuum system, safety system, and the two-beam chamber area is elements of the diagnostics systems. protected by beam tubes, which can We replicated Nova designs for be evacuated or gas-filled. The beam required new components wherever tubes serve two purposes. First, they possible instead of making costly provide insulation from the air redesigns to achieve only nominal turbulence in the class-10 000 clean performance improvements. rooms: air turbulence can seriously We moved the Novette space frame degrade the quality and focusing to a new location in ihe laser bay characteristics of the high-intensity previously occupied by Shiva to serve laser beams. Second, they preserve the as the support frame for the target quality of the laser light over long chamber and the optics located in the paths. Recent experiments have beam lines near the chamber. The verified that high-intensity light Shiva space frame, which still propagating through long paths of occupied much of the Shiva laser bay, nitrogen can be scattered into different was removed. The Novette frame was wavelengths through a process called cut into seven segments, installed in stimulated Raman scattering. This the Shiva laser bay, and rewelded scattering reduces the efficiency of our using low-distortion welding frequency-conversion arrays, resulting techniques. We added mirror supports in less focusable power on target at on this frame and in the Nova the desired wavelengths. Since switchyard to control beam divergence nitrogen makes up approximately and pointing. These mirrors and their 80% of our normal atmosphere, we electronically controlled gimbals are designed the beam tubes to be able to identical to those designed for Nova. hold a vacuum or to be filled with a We diverted two of Nova's beam nonscattering gas like aigon or neon, lines from the optical switchyard area as necessary. into what once was the Shiva laser bay. This eliminated the need for onstruction and additional construction and left room Activation for a third and larger chamber in the C In early September 1984, we Shiva target area, should a future began work to design and install the need arise. We chose two Nova beam auxiliary target chamber for the Nova lines whose optical path lengths to the laser facility. We had to meet a very ten-beam chamber corresponded with tight schedule (we wanted to begin the path lengths to the two-beam x-ray laser experiments with the Nova chamber. Thus the light pulses would laser in September 1985), and we arrive, in synchronization, at the could not delay completion of the center of either chamber without time- Nova facility itself (planned for consuming adjustment of path length December 1984) or the experimental when operations were switched from program that was to start soon one chamber to the other. thereafter. In the two-beam chamber area, We were able to complete the one beam line is reflected off a addition of the two-beam chamber, motorized mirror into one end of the while meeting these criteria, through two-beam chamber, after passing close coordination with the activities through alignment, diagnostics, of the Nova construction and frequency-conversion, and final- activation personnel. In many cases, focusing optics. The other beam line, engineering and fabrication were which enters ihe area at an elevated accomplished by the same workers level, is lowered to the correct level who were activating Nova. INERTIAL FUSION

Procurement and facility using the two-beam target chamber. preparation were the major activities These experiments, interwoven with from September to April 1985. In experiments using the ten-beam April, the area was operated as a chamber, demonstrated the class-10 000 clean room as we started effectiveness of this dual-target- installing the chamber, opto­ chamber configuration for Nova, mechanical assemblies, and finally the permitting us to prepare for optics themselves. On July 31. 1985, the relocation of the target chamber from the Novette laser area to the Shiva laser bay was completed. We marked this accomplishment at 9:50 p.m. that evening by firing the two beams simultaneously into the chamber onto a nickel plate target. (The auxiliary target chamber and two-beam configuration are shown in Fig. 1.) The two beams delivered 1.5 and 1.3 kj, respectively, of second- harmonic light in a 400-ps, temporally Gaussian pulse with approximately a 1-cm-long line focus. X-ray images were captured by two pinhole cameras mounted on the target chamber. Data were also taken by two x-ray spectrometers. The target surface was bumed as expected, demonstrating good beam alignment. In this experiment, we acquired incident laser diagnostic data related to beam energy and temporal history. We also demonstrated that the use of incoherent illumination for target viewing during alignment was a major improvement over previous, coherent illumination strategies. During August 1985, we concentrated on activating the chamber for target experiments, including the continuation of the soft- x-ray laser experiments. We also activated major x-ray laser diagnostics and completed analysis of the focal spot characteristics, system performance, and diagnostic instruments. In September, we began the next series of soft-x-ray laser experiments, which continued throughout the remainder of the year. These initial experiments yielded significant advances in the soft-x-ray laser experimental program (see the article on p. 34). Fig. 1 From September through December Artist's rendering of the Nova laser facility with the auxiliary target chamber installed, at left, in the 1985, 102 experiments were conducted Shiva iaser bay (a) ana photograph of the two-beam auxiliary target chamber (b).

71 experiments in one chamber while of the Nova ten-beam chamber. We conducting experiments in the other. significantly improved operations and maintenance, however, by mounting inal Focusing these elements with slides connected Configuration to precision horizontal rails (as on FThe final train of optics through Novette). Figure 2a illustrates the nine which the beams pass on their way to major elements in this final optical the target was designed to be flexible train as the beams approach the target and reconfigurable, In general, the chamber. These elements are: configuration of the optical train is • An optional, full-aperture, analogous to that used for each beam infrared calorimeter to measure full beam energy. The calorimeter is removed for target experiments. Fig. 2 • A frequency-conversion array of Major optical and diagnostic assemblies in the Full-aperture KDP crystals. final optical train aa the laser beam ap­ (74-cm-diameter) • An optional beam dump to proaches the target chamber (a). Diagram of calorimeter absorb unwanted infrared light during the cylindrical lenses that provide the Nne KDP array experiments using second-harmonic focus

* ' I

KDP array Counter-rotating Fused silica Target cylindrical lenses focusing lens chamber center

72 • An optional 44-cm calorimeter Notes and Reference* that can be placed inside the target 1 A mini of Ihe desj^n of thi- \ma J,wr chamber to measure actual energy -vstem appear- in W W Simmon1- rf ,1/ reaching the center of the vessel. This kngiimTinf; M'si^n of (he \m,i l,iicf /iwOfi l-.hih'v fur fnt'rtu! C'onfim'tTH'if. is used to calibrate the incident /usjoii. I awrence r.i\t*nT»tifi* \jtional diagnostic sensor package. I abur.uon. Kepi COM KHO-ld (January 2?. I9S2). onclusion 2 To t unvrr! Ihe freiiiii'r.cy or the KT-.IT li^rv. The addition of the two- each last-r beam pa-.-*", through ,jn arra\ of K'Ur^MitjvMum dihulixttrn phu-phatr) beam chamber to Nova has C crystal windows (hot convert tht,' proven to be very valuable. Effective fundamenlal infrared wavelength to either operation of Nova is accomplished by green lij^ht {second harmonic] or interleaving major setup times in one ultraviolet light flhird harmiinir) just chamber with experiments in the before ihe beam enters the tar>;el chamber We choose ihe thickness of these crystal other. We have successfully performed windows jri .irdmg to ihi' color experiments in both chambers on the (frequency) tn which we- wish to optimi/e same day.3 With this second target Target-physics ,mJ weapon-physics chamber, we can perform soft-x-ray experiment*, p.'nerath perform better wiih laser and other experiments without ultraviolet li>;h: Accordingly ihi- KDr array- used on the ten-beam chamber ire significantly impacting target-physics optimized in ihickneis to p*.e m^wrnvm and weapon-physics experiments in conversion ol infra red lo ultraviolet li>;ht the larger ten-beam chamber. The wav laser experiment-, for which the One of the most significant results two-beam chamber was optimized, were pbnni'd HIT ^reen li>*h!. Therefore, the of this effort was that we were able to KDP arrays for this chamber are currently satisfy important experimental needs optimized for green li&ht conversion, by making more productive use of a although thee stil] provide reasonably major available facility through a cost- efficient conversion to ultraviolet light effective extension. U y The ten-beam target chamber and recent experiments are discussed in more derail :n Key Words: teHT—\oea. \ovrtte. Shn-a, way: :ht* /.a*er Program Annual /fe/wf ftp. target chamber—auxiliary. *-eciind. ten-beam, Ijwtence Lnermore National Laboratory. two-beam. Kepi UCRI.-5002I-K5 (in pn-ss)