Progress in Laser Fusion

• Progress 1n Laser Fusion Infrared laser beams, converted to shorter wavelengths by optical crystals, may heat heavy hydrogen to 100 million degrees Celsius. Recent work suggests the technique could lead to useful fusion power by R. Stephen Craxton, Robert 1. McCrory and John M. Soures n the past year four of the world's largest laser systems, the NOVA la­ ser at the Lawrence Livermore National Laboratory in Berkeley, the GEKKO-XII laser at the University of Osaka in Japan, the PHEBUS laser at Li­ meil, France, and the OMEGA laser in our laboratory at the University of Rochester, have come into operation at visible and ultraviolet wavelengths. At the Los Alamos National Labora­ tory another kind of laser, operating at a slightly shorter ultraviolet wave­ length, is under development. Many large laser systems play an undisputed role in weapons research and develop­ ment. Nevertheless, the lines of scien­ tific investigation opened up by all these instruments make it possible to be guardedly optimistic about an out­ standing peacetime benefit: the scien­ tific feasibility of harnessing fusion power by laser may well be demon­ strated in the next 10 years. Major advances in two disciplines are primarily responsible for this optimism. In optics a phenomenon called harmonic-frequency conversion has been exploited to generate high­ powered laser beams that have a much shorter wavelength than was previous­ ly possible. The advance is significant because laser energy can be trans­ ferred to a fusion fuel pellet much more efficiently at short wavelengths than it can at long wavelengths. In computer modeling the simula­ tion of fusion reactions initiated by la­ sers in fuel pellets has become increas­ ingly detailed and realistic. Simula­ tions now make it possible to specify quite accurately the characteristics of fuel pellets and incident laser beams that are needed to generate economi­ cally viable fusion power. Moreover, remarkable instrumentation can now trace the evolution of the fusion reac­ tions to within a hundredth of a nano­ second. (A nanosecond is a billionth of a second.) The controlled release of the energy I 68 derived from the fusion of atomic nu­ clei has been one of the fondest hopes of science since the potential benefits of the process were recognized and serious work on it was begun nearly 40 years ago. It is well known that a mixture of deuterium and tritium, the heavy isotopes of hydrogen, makes the most suitable fuel for fusion. The ba­ sic scientific objective is to heat ions of deuterium and tritium to sufficiently high temperatures and maintain the temperatures long enough to realize a net energy gain from the fusion of the ions. In each fusion reaction a deuteri­ um ion and a tritium ion give rise to an alpha particle, or helium nucleus, and a neutron; the total kinetic energy re­ leased, which is shared by the two re­ action products, is 17.6 million elec­ tron volts. In the long term, fusion between deuterium ions alone could provide a practically limitless source of energy, since vast quantities of deuterium are available from seawater. For example, the energy from the fusion of the deu­ terium in a pool of water 100 feet on each side and seven feet deep could satisfy the electrical energy needs of the city of Rochester for a year. Laser Fusion The best-known scheme for creating the conditions needed for fusion has been to heat a plasma, a hot, electrical­ ly neutral gas of ions and free elec, trons, that is confined by a strong mag­ netic field [see "The Engineering of Magnetic Fusion Reactors," by Rob­ ert W. Conn; SCIENTIFIC AMERICAN, October, 1983]. In the early 1970's, however, workers in the U.S. began experiments on a radically different method of heating and confining a plasma, one that exploits the extraor­ dinary heating potential of the laser [see "Fusion by Laser," by Moshe J. Lubin and Arthur P. Fraas, SCIENTIF­ IC AMERICAN, June, 197 1, and "Fusion © 1986 SCIENTIFIC AMERICAN, INC Power by Laser Implosion," by John L. Emmett, John Nuckolls and Lowell Wood, SCIENTIFIC AMERICAN, June, 1974]. A beam or pulse of laser light is split into several smaller beams of equal intensity. The split beams are amplified in energy and subsequently brought back together by a system of mirrors and lenses; the beams are thereby focused on a small region from different directions. A charge of deuterium and tritium fuel is encased in a spherical shell a few millimeters in diameter, made of plastic, glass or some other material, and the resulting fuel pellet is placed at the intersection of the beams; the pellet is thus uni­ formly illuminated. The laser pulse almost instantly ion­ izes the atoms in the outermost layer of the pellet, but the material inside a certain critical radius is opaque to the laser energy. Incident energy is conse­ quently absorbed in a dense layer of plasma that surrounds the deuterium­ tritium fuel. The heated layer of plas­ ma expands and ablates, or becomes explosively torn free, from the rest of the pellet; the velocity of the ablated plasma is typically 1,000 kilometers per second. An equal and opposite force accelerates the material inside the ablation layer inward, in accor­ dance with Newton's third law, as if it were a rocket propelled by the plasma escaping all around it. The concentric implosive force is sufficient to acceler­ ate the remaining shell to a velocity of several hundred kilometers per second in a billionth of a second. The radius of the fuel is compressed as much as 50 times, and the resulting high tempera­ ture and high density of the fuel cause it to fuse. In a working fusion reactor the neu­ trons to which the reaction gives rise would escape from the fuel at high ve­ locities and deposit their energy in the surrounding reactor chamber. In one design a layer within the wall of the re­ actor chamber would hold a fluid such TARGET AREA of the OMEGA laser system, operated in the au­ thors' laboratory at the University of Rochester, is shown in the photograph. The system is a neodymium-glass laser converted to a shorter wavelength in order to study the properties of fuel pellets for laser fusion. The laser first emits a single beam of infrared light at a wavelength of one micron. The beam is then amplified and split into 24'beams. After further amplification each of the 24 beams is passed through a cell holding two crystals of potassium dihydrogen • phosphate (KDP), where it is converted into an ultraviolet beam at three times the input fr equency. Mirrors and lenses guide and focus the beams into the target chamber, the complex spherical device at the center of the photograph. The target chamber is evacuated, and the fuel pellet is suspended at its center; the 24 focused laser beams converge on the pellet and deposit their energy there. Instruments for monitoring the effects of the laser energy on the pellet can be seen protruding from the upper hemisphere of the target chamber. © 1986 SCIENTIFIC AMERICAN, INC 69 as lithium, which is known as a moder­ ator. The neutrons would collide with the fluid and slow down, giving up their kinetic energy to the fluid in the form of heat. The fluid would circu­ late to a heat exchanger, which would then transfer the heat to pressurized, circulating steam, and electric power would be generated by steam-driven turbines in the usual way. The entire process is known as laser fusion. Although we shall limit our discus­ sion to laser fusion, we should note that particle-beam accelerators can also drive the compression of a fuel pellet. At the Sandia National Labo­ ratories in Albuquerque construction has recently been completed 011 an ac­ celerator called PBFA " (Particle Beam Fusion Accelerator), which will ener­ gize fuel pellets with beams of high-en- INCOMING LASER BEAM ABLATION SURFACE REFRACTED LASER RAY QUARTER-CRITICAL SURFACE IMPLODING PLASTIC SHELL CRITICAL-DENSI T Y SURFACE LASER FUSION takes place in a fuel pellet, typically a small plas­ tic shell about a millimeter in diameter that holds a mixture of deu­ terium and tritium fuel. The shell is irradiated uniformly from many directions by overlapping laser beams, and its outer portion is vaporized to form a plasma. The laser beams penetrate the plasma only up to the critical-density surface and deposit much of their energy in its vicinity. One mechanism for the energy absorption, called collisional absorption, is particularly effective at short laser wavelengths. Electrons oscillating in the electric field of the laser heat the plasma through collisions with one another and with ions in the plasma. The thermal energy of the heated plasma is then con- ducted inward to the ablation surface, the boundary between the hot plasma and the dense shell. The plasma outside the ablation surface explodes away from the rest of the shell like the exhaust plume of a rocket. The reaction force causes the shell to be driven inward to­ ward the center of the fuel pellet. The implosion compresses the fuel into a dense core and heats it; thermonuclear reactions can then take place. For long-wavelength lasers a substantial fraction of the laser energy is deposited into unwanted, highly energetic supra­ thermal electrons. Such electrons can penetrate the core and heat the fuel in the initial stages of the implosion, thereby precluding the high compression needed in the core to achieve fusion ignition. 70 © 1986 SCIENTIFIC AMERICAN, INC ergy lithium ions. The general term for all such schemes of generating fusion in a pellet by compressing it, whether by laser, particle beam or some oth­ er energy source, is "inertial-confine­ ment fusion." Frequency-converted Lasers Until quite recently lasers powerful enough to drive fusion reactions had been built only for emission wave­ lengths in the infrared. Most of the ex­ perimental work to develop laser fu­ sion has been conducted with two kinds of infrared laser: the solid-state, neodymium-doped glass laser, whose wavelength is one micron, and the car­ bon dioxide gas laser, whose wave­ length is 10 microns. By the end of the 1970's, however, extensive experi­ ments and elaborate computer simula­ tions had identified a fundamental dif­ ficulty. For lasers emitting at the in­ tensities needed for fusion and at wavelengths of one micron or longer, less than half of the laser energy is ab­ sorbed by the fuel pellet. Furthermore, of the energy that is absorbed a sub­ stantial fraction gets carried away by so-called suprathermal electrons. Such electrons pass freely through the fuel pellet and heat the fuel before it can be compressed. The premature heating precludes achieving the high densities needed for fusion. Investigators quickly realized that the unwanted preheating of the fuel might be avoided if high-power lasers having wavelengths shorter than the infrared could be built. The apparent impasse posed by the unavailability of such lasers was sidestepped by an interesting optical trick. By carefully growing perfect crystals and then cut­ ting each one at a precisely calculated angle with respect to the orientation of its crystal lattice, devices can be made that generate the higher harmonics of the incident beam. The frequency of the harmonics is two, three or a high­ er integer times the frequency of the input wave, depending on the number of crystals employed and their orien­ tation within each device; the wave­ length of the corresponding higher­ harmonic beams is shortened by the same integral factor. For example, if long-wavelength in­ frared laser light is passed through one crystal, a large fraction of the light energy can emerge as shorter-wave­ length green light at twice the frequen­ cy of the input. If the green light and the residual infrared light from the first crystal are passed through a sec­ ond crystal, they can be combined into a beam of even shorter-wavelength ul­ traviolet light with three times the fre- • 100 l=' z w u a: w eo. z 0 i= V4MICRON 80 V3MICRON 60 0a: 0 en CD « >(!) a: w z w 40 20 o� ______________ 10" � �� ____ ________ __ __ ____ __ ____ 1014 10" � 10'6 LASER INTENSIT Y (WATTS PER SQUARE CENTIMETER) ABSORPTION EXPERIMENTS at different laser wavelengths show that the energy ab­ sorbed by a fuel pellet increases sig nificantly as the wavelength of the laser decreases to less than one micron. The effect is particularly strong in the intensity range of 10" to 10" watts per square centimeter, which is the range required for generating fusion in a reactor with laser irradiation. For such wavelengths the high absorption at low intensity and the decline in absorption at higher intensities are characteristic of collisional absorption, which leads to the desired heating of the plasma surrounding the pellet. The fraction of the inci­ dent energy absorbed by unwanted suprathermal electrons increases with laser wavelength. quency (a third of the wavelength) of the original infrared beam. In one experiment at the Ecole Pol­ ytechnique in France a one-micron beam from a neodymium-glass laser was passed through two crystals of po­ tassium dihydrogen phosphate (KDP) in succession; the wavelength of the beam was reduced by a factor of two in each crystal, and the total reduction was therefore a factor of four. A target pellet absorbed almost all the energy of the frequency-converted beam, and suprathermal electrons were not de­ tected. In such early experiments only a small part of the incident light was converted into light of higher frequen­ cy, but the results were still very en­ couraging. Investigators began direct­ ing greater attention to the physics of harmonic-frequency conversion and to methods for improving the conver­ sion efficiency of existing neodymi­ um-glass lasers. Temperature and Density To appreciate the need for the ex­ traordinary scientific efforts being de­ voted to frequency conversion and other aspects of laser-fusion technolo­ gy one must understand the physical environment required if laser fusion is to generate economically useful ener­ gy. The temperature of the fuel in each pellet must reach almost 100 million degrees Celsius, several times the tem- perature at the center of the sun. Si­ multaneously the fuel must be com­ pressed to a density typical of what is found in the center of the sun: more than 1,000 times its ordinary solid or liquid density of .2 gram per cubic cen­ timeter. The pressures corresponding to such a temperature and density are immense; indeed, pressures on the order of several billion atmospheres have already been generated in laser­ driven implosions. Furthermore, mac­ roscopic quantities of fuel must be brought to such conditions. The cre­ ation of such an environment not only would have immediate application to fusion power but also would open the possibility of doing experiments of di­ rect relevance to astrophysics in the terrestrial laboratory. The need to heat fusion fuels derives from the basic energetics of nuclear in­ teractions. At a very short range the protons and neutrons that make up atomic nuclei are attracted to one another by the strong nuclear force. Many systems of nuclei can therefore increase their binding energy if the nu­ clei fuse. The total mass of the reac­ tion products is slightly less than the sum of the masses of the fusing nuclei. Thus, in accord with Einstein'S equiva­ lence between energy and mass, the mass lost in the reaction is converted into energy, which shows up as the ki­ netic energy of the reaction products. There is an important energy barrier 71 © 1986 SCIENTIFIC AMERICAN, INC to this path for reducing potential en­ ergy. At temperatures less than about 10 million degrees C. the mutual elec­ trostatic repulsion of the positively charged protons in the nuclei keeps the nuclei from coming close enough to "feel" the attraction of the strong force. Only within a distance of about 2 X 10-13 centimeter-comparable to the radius of the nucleus itself-does the strong force become dominant. To approach within this range two nuclei must be brought together with consid­ erable kinetic energy; in other words, their temperature must be raised sub­ stantially. An important reason for the selection of deuterium and tritium ions as fusion fuels is that each ion car­ ries only one positive charge, and that small charge minimizes the mutual re­ pulsion of the fuels. High compression is not a theoreti­ cal prerequisite for fusion, but it en­ hances the efficiency of the fusion re- actions, and it is essential in a practical reactor design. The laser input energy needed to fuse deuterium and tritium efficiently at ordinary densities is well beyond the capacity of present laser technology. When the fuel in the pellet reaches its peak compression, the ki­ netic energy of the imploding material is converted into heat, and the confine­ ment of that heat to the small com­ pressed region raises the temperature of the fuel. Thermonuclear burning begins in a small "spark plug" region at the center of the fuel. Four-fifths of the energy released by the burn­ ing emerges as the kinetic energy of neutrons, which pass through the sur­ rounding plasma and deposit their en­ ergy into a fluid circulating in the reac­ tor chamber or into a moderator in the reactor wall. The other 20 percent of the energy released by the fusion reactions is car­ ried away by alpha particles. Because z z OPTIC AXIS OPTIC AXIS DIREC TION OF LASER PROPAG ATION II I I I \ / / / / / / x \ /\\ \\ v- \ /1 \' : I ,,\ \ "" '" \\1 \ �I , I' 0 , lit \ \ \ \I 1\ " I, I I I I I I I / \ \� I e " �\ \ / \ / / / )- " \\ \\ '- \\ \\ \\ " \ " \ \ \ I \ , \ \\ \ \ \ \ \ \ 'I /\ I 1..--"- \ \ \ \ "f�".Lf\ -- / -- - - 7 - \ ,If /y \ --- x DIRECTION OF LASER PROPAG ATION CRYSTALS for generating higher harmonics of the incident laser frequency are sliced out of a crystal block as circular disks with a predetermined orientation to the three crystallographic axes x, y and z (le!t). The crystal is symmetric under 90-degree rotations about the z axis, which is known as the optic axis. The laser is propagated through the disk at a right angle to its surface, making an angle called the phasematching angle with the optic axis. The phase-matching angle is determined by the way the crystal disk is cut (above). A light wave entering the crystal with an arbitrary polarization splits into two waves. The first wave is the ordinary, or 0, wave polarized at a right angle to the optic axis and along, say, the y axis of the crystal. The second wave is the extraordinary, or e, wave polarized in the (x-z) plane and perpendicular to the 0 axis of the crystal disk. In the absence of nonlinear effects the two waves travel independently through the crystal at slightly different speeds, which are determined by their slightly different refractive indexes. 72 © 1986 SCIENTIFIC AMERICAN, INC they are charged particles, they are slowed by the fuel much mqre effec­ tively than the neutrons are. They col­ lide with the surrounding fuel and give up their kinetic energy over a distance inversely proportional to the fuel den­ sity. A sizable fraction of the fusion energy can thus be deposited in the cooler layers of the compressed fuel if the fuel is dense enough and its radius is large enough to stop the alpha parti­ cles. Such partial absorption of the fu­ sion energy in the fuel is critical to the efficient propagation of thermonucle­ ar burning; it is called bootstrap heat­ ing. When the energy deposited in the fuel by the alpha particles exceeds the energy needed to compress the fuel, the pellet is said to have ignited. If the attainable value of the compressed fuel density is known, the core radius, the fuel mass and the laser energy needed for ignition can all be deter­ mined. Achieving ignition would be a milestone in demonstrating the scien­ tific feasibility of laser fusion. There is a small price to be paid for high compression: soon after fusion begins it must cease because the high internal pressure of the compressed fuel causes it to fly apart. The higher the compression is, the faster the fuel disassembles. Nevertheless, the high reaction rate at high compression more than compensates for the limit­ ed burning time. For example, other things being held constant, a tenfold compression of the radius of the fuel increases the reaction rate by a factor of 1,000, whereas the duration of the burning is reduced by a factor of only 10. It should be noted that in magnet­ ic fusion devices the density of the fuel is 10 billion times lower than it is in the burning core of the pellet. The de­ sign must compensate for the low den­ sity by confining the fuel for 10 bil­ lion times as long. Collisional Energy Absorption Given the need for high temperature and high density of the fuel, which of the many physical processes relevant to laser fusion are the most important to understand and control? It turns out that the most critical processes depend on details of the interaction of the la­ ser light with the plasma corona, or at­ mosphere, surrounding the fuel pellet. For example, the transfer of energy from the laser to the ablation layer be­ gins in the corona. Furthermore, the degree to which the fuel is preheated by suprathermal electrons, prior to the implosion, is intimately related to the. plasma physics of the corona. After the first incident laser irradia­ tion ionizes the outer surface of the pellet, the resulting plasma corona be­ gins to expand outward. The corona is dense near the surface of the pellet and becomes rarefied at larger radii. The rapidly oscillating electric field of the subsequent incoming laser light causes rapid oscillations of the light­ est charged particles in the plasma, namely the electrons. A current is therefore set up in the plasma, oscil­ lating at the frequency of the driving electric field and proportional to the density of the electrons in the plasma. In the rarefied outer regions of the corona the current is small, but in ·the inner regions, where the electron den­ sity is greater, the current becomes larger. At what is called the critical ra­ dius the current becomes large enough to shield the plasma at smaller radii from further penetration of the incom­ ing electric field. The oscillating elec­ tric current acts as an antenna that broadcasts a second, outgoing electro­ magnetic wave, equal in frequency to the incoming wave but carrying en­ ergy away from the target. The elec­ tron density at the depth of maximum penetration of the incoming wave is called the critical density, and it is in­ versely proportional to the square of the laser wavelength. For example, a threefold reduction of the laser wave­ length would give rise to a ninefold increase in the critical density. Much of the absorption of the laser energy takes place in the vicinity of the critical-density surface. The oscillat­ ing electrons collide with ions in the plasma as well as with one another, thereby transferring part of their ener­ gy to the plasma as the energy of ran­ dom motion, or heat. This mechanism for energy transfer is more efficient for a short-wavelength laser than it is for a long-wavelength laser, since the short-wavelength beam can penetrate to higher electron densities where col­ lisions are more frequent. The heat en­ ergy is conducted inward to the abla­ tion surface, the boundary between the exploding and imploding regions of the pellet. Collisional absorption ac­ counts for virtually all the absorbed energy that contributes to the implo­ sive compression of the pellet. Suprathermal Electrons There are several other mechanisms for laser-energy absorption that do not contribute to the heating or compres­ sion of the fuel but must be under­ stood because they give rise to un­ wanted suprathermal electrons. As the name implies, the energy of a supra­ thermal electron is significantly higher than the average energy of the thermal electrons in the plasma. Such excess z OP T IC AXIS DIRECTION OF LASER PROPAGATION PHASE-MATCHING ANGLE FOR FREQUENCY TRIPLING PHASE-MATCHING ANGLE FOR FREQUENCY DOUBLING x ORIENTATION OF DISK SLICE with respect to the crystallographic axes is determined by a simple geometric construction. The red arcs represent the input frequency of the infra­ red laser, and the green and blue arcs respectively represent the second and third harmonic frequencies. The refractive indexes of the 0 waves vary with wavelength but not with a change in the orientation of the slice with respect to the z axis. Hence the refractive index of the infrared 0 wave is shown as a red circular are, and the indexes of the green and ultraviolet 0 waves are shown as green and blue circular arcs. The refractive indexes of the infrared, green and ultraviolet e waves do vary with the angle the slice makes with the z axis; the variations are plotted as red, green and blue broken elliptical arcs. The propaga­ tion direction for frequency doubling must be chosen in such a way that the refractive index of the green e wave is equal to the average of the refractive indexes of the infrared 0 and e waves. For frequency tripling the refractive index of the ultraviolet e wave must equal one­ third of the index of the infrared e wave plus two-thirds of the index of the green 0 wave. energy can be acquired only if the elec­ tron is accelerated by a large electric field for a sufficient time. The electric field of the laser, although intense, changes direction so often that the ve­ locity of the electrons oscillating in the field is relatively small. Hence in most experimental conditions the electric field of the laser does not lead directly to suprathermal electrons. Large electric fields can nonetheless arise in waves known as plasmons, which travel through the plasma. Plas­ mons are somewhat like sound waves in that they propagate as compressions and rarefactions of particles in the di­ rection of the wave motion. Unlike sound waves, however, plasmons do not affect all the particles in the medi­ um: the ions in the plasma are station­ ary and only the electrons move. Re­ gions of net positive charge are there­ by created in the rarefied electronic regions, and regions of net negative charge are created in the compressed regions. A large electric field is formed between a rarefied region and a com­ pressed region, and there is a strong at­ tractive force between the two. Electrons that happen to be moving in the same direction as the plasmon and at roughly the same speed become trapped in the plasmon somewhat like a surfer riding a wave. Since they move with the same speed as the wave, they experience its electric field as a nonoscillating field. The field acceler­ ates them to suprathermal velocities, which enable them to escape the plas­ ma layer before collisions can slow them down. Plasmons can be generated in many ways. One important mechanism is called resonance absorption. As a laser ray penetrates the plasma, much of the energy of its oscillating electric field is given up to surrounding electrons. Near the critical-density surface, how­ ever, the natural oscillating frequency of plasmons equals the laser frequen73 © 1986 SCIENTIFIC AMERICAN, INC ELECTRIC FIELD OF INFRARED INPUT DOUBLER CRYSTAL e ONE GREEN PHOTON TRIPLER CRYSTAL o DIRECTION OF LASER PROPAG ATION e POLARIZATION ANGLE 35 DEGREES THREE INFRARED PHOTONS ONE-TO-TWO MIX ELECTRIC FIELD OF INFRARED INPUT DOUBLER CRYSTAL e GREEN LIGHT GENERATED IN FIRST CRYSTAL � ONE ULTRAVIOLET PHOTON DOUBLER CRYSTAL o DIRECTION OF LASER PROPAG ATION � POLARIZATION ANGLE 45 DEGREES GREEN LIGHT GENERATED IN SECOND CRYSTAL TWO INFRARED PHOTONS ONE-TO-ONE MIX ROCHESTER FREQUENCY-TRIPLING SCHEME (top) em­ ploys two crystals whose 0 and e axes are mutually perpendicular. The incident infrared laser beam is polarized at an angle of 35 de­ grees to the 0 axis of the first crystal. This ensures that two-thirds of the photons incident on the first crystal are aligned with its 0 axis and one-third are aligned with its e axis. One 0 photon com­ bines with an e photon to give an e photon at the second harmonic (greell), or at twice the frequency of the incident light. The green e photon combines in turn in the tripler crystal with the remaining infrared photon to give one ultraviolet photon at the third harmonic (blue). In the authors' laboratory the technique has been shown to cy. The energy that reaches this depth in the plasma can drive plasmons res­ onantly to large amplitudes, much as a child on a swing can go progressive­ ly higher by pumping in synchrony with the swing's natural motion. The energy that is pumped into resonantly driven plasmons is eventually released as the kinetic energy of suprathermal electrons. be 80 percent efficient, and it is now employed in the NOVA laser at the Lawrence Livermore National Laboratory. A simple reorienta­ tion of the two crystals (bottom) also makes it possible to generate the second harmonic in the NOVA laser with improved efficiency. The input beam is polarized in such a way that the numbers of infrared 0 and e photons are equal. The doubling process is less than 100 percent efficient, and some residual infrared photons remain unconverted after passing through the first crystal. When the sec­ ond crystal is tilted through a small angle from its position for fre­ quency tripling, it becomes a doubling crystal, and residual infra­ red photons have a second opportunity to be converted to green. There are two further important mechanisms whereby plasmons can form and generate suprathermal elec­ trons. Although both mechanisms ap­ pear to be less significant causes of preheating than resonance absorption, it is worthwhile to explain them brief­ ly. Both of them arise out of a phe­ nomenon called three-wave mixing, which also plays a major role in har- X-RAY IMAGES can determine the uniformity of laser irradiation on a fuel pellet. The color-enhanced image at the left shows the X­ ray emission when all 24 beams of the ultraviolet OMEGA laser are focused onto small areas of the pellet. Such illumination does not lead to symmetric compression, but it is useful for testing the point­ ing and focusing of the laser beams. The other images are from imploding pellets irradiated with overlapping beams. In the middle monic-frequency conversion. In three­ wave mixing two waves can interact through the matter in a plasma or a crystalline solid to produce a third wave. In general the mixing is stron­ gest when the amplitudes of the inter­ acting waves are large. The frequency of the third wave equals the sum of the frequencies of the two input waves. The direction of the process is also re- is a large pellet with a thin glass shell and at the right is a smaller pellet with a thick glass shell; both pellets are filled with deuterium and tritium gas. The red parts of each image are the regions of highest X-ray emission from the compressed part of the shell. The thick shell maintains the fuel at a lower temperature than the thin shell does, and it also generates a higher compression of the fuel. The X-ray emission extends just beyond the initial pellet surface. 74 © 1986 SCIENTIFIC AMERICAN, INC versible: from a "product" wave two new "factor" waves can emerge. The first mechanism for the forma­ tion of plasmons is known as the two­ plasmon instability. Here the laser beam functions as the product wave, and it splits into two plasmons, which function as the two factor waves. The frequency of each plasmon is half the frequency of the incoming beam. Since the critical density of the plasma electrons varies inversely with the square of the input wavelength, the two plasmons are driven resonantly when the electron density in the plas­ ma is a quarter of the critical density for the input laser. Hence the two-plas­ mon instability arises near the so­ called quarter-critical surface [see il­ tests of these effects are needed for the larger pellets required in a commercial laser-fusion reactor. Thus empirical study to date has confirmed the signifi­ cant advantages of short-wavelength lasers for fusion. Frequency-Conversion Crystals The desire to investigate how la­ ser energy at submicron wavelengths can be exploited to compress fuel pel­ lets has led to the recent interest in frequency-converted lasers. Since fre­ quency conversion can readily shorten the wavelength of a laser to a half or a third of its ordinary emission value, the wavelength range of interest is ex­ perimentally accessible with existing neodymium-glass lasers. Optical frequency conversion may become less critical as new kinds of la­ sers appear. For example, the krypton­ fluoride laser, a gas laser under devel­ opment at the Los Alamos National Laboratory and elsewhere, can also lustration on page 70]. The second mechanism for plasmon formation is called Raman scattering; again the laser beam is the product wave, and its factor waves are a plas­ mon and a reflected light wave. Ra­ man scattering can take place at the quarter-critical surface and in more rarefied regions of the plasma. The plasmons created by the two-plasmon instability and by Raman scattering generate suprathermal electrons in much the same way as they do in reso­ nance absorption. When the first suprathermal elec­ trons escape from the corona, they leave a net positive charge on the pel­ let. The charge attracts later, outward­ moving suprathermal electrons back toward the pellet, whereupon they overshoot their original positions and pass on into the pellet core. Some of the most energetic electrons may oscil­ late several times through the pellet before collisions with the fuel finally slow them; the transfer of collisional energy to the fuel preheats the fuel. The attractive electrostatic force be­ tween the suprathermal electrons and the ions in the corona can also lead to the outward acceleration of highly en­ ergetic ions. The loss of energetic ions from the ablation layer also drains away laser energy that might other­ wise drive the implosion. Experiments show that for the range of beam intensities relevant to laser fusion-generally from 101 4 to 1015 watts per square centimeter-colli­ sions are weak at infrared wavelengths and resonance absorption is dominant. For wavelengths shorter than .5 mi­ cron, however, a variety of experimen­ tal results suggest that collisional ab­ sorption is the most important absorp­ tion mechanism. Moreover, at such wavelengths neither the two-plasmon instability nor Raman scattering have been found to preheat the fuel signifi­ cantly, although further experimental o 5 . TIME (10-9 SECOND) GLASS SHELL OF PELLET r (/) z 0 � a: () 0 0 Ci:J 1 VIEWING REGION OF IMAGING STREAK CAMERA © 1986 SCIENTIFIC AMERICAN, INC 1.0 1.5 IMPLOSION HISTORY of a spherical, glass fuel pellet uniformly illuminated by the OMEGA laser is captured in this color­ enhanced photograph. An X-ray pinhole camera forms an image of the pellet similar to the ones in the bottom illustration on the opposite page, and a mask with a slit is placed over the X-ray image (left). A streak camera then displaces the part of the image seen through the slit a distance to the right that is proportional to the time elapsed dur­ ing the implosion. The resultant image (up­ per image) thus portrays the implosion in both space and time. Soon after the laser begins illuminating the pellet, X rays are emitted from the plasma near its initial sur­ face. As the pellet is compressed the region emitting X rays moves closer to its center, until a bright flash of X rays is visible at peak compression, about one nanosecond af­ ter the implosion begins. The speed of the imploding shell is about 200 miles a second. 75 deliver the high energies needed for fu­ sion implosion; its emission wave­ length is only .25 micron. Yet although the krypton-fluoride laser is widely· considered to be a promising candi­ date for a reactor laser, questions re­ lated to the efficiency with which it can deliver the short pulses of ener­ gy needed in a reactor are still under study. The neodymium-glass laser re­ mains the major research tool for the study of short-wavelength, laser­ fusion implosions. Many kinds of crystal have been in­ corporated into small laser systems for converting laser light to its higher har­ monic frequencies. The crystal KDP, however, is the only one that has so far been grown large enough for current work. KDP crystals are grown from solution at a rate of only a few centi­ meters per month. The largest avail­ able crystals of acceptable quality are more than 30 centimeters in diameter and take as long as a year to grow. Cir­ cular slices about one centimeter thick are cut from the crystal, polished and mounted in the path of a laser beam. One crystal can generate the second harmonic, a wave at twice the frequen- cy of the fundamental beam. A second crystal mounted in the path of the first output beam can then mix the generat­ ed second harmonic with the residual fundamental to produce the third har­ monic [see top illustration on page 74], or the crystal can generate the fourth harmonic by a second doubling proc­ ess. In principle there is no upper limit to the frequency that could be generat-I ed by successive stages of doubling, but unfortunately suitable crystals transparent to wavelengths shorter than .2 micron are not available. Harmonic Generation The capacity of some crystals to generate higher harmonic frequencies of a laser beam was discovered soon after the invention of the laser. Investi­ gators at the University of Michigan found that a laser beam focused into a crystal of quartz emerged from the crystal with some light at the second­ harmonic frequency. As we mentioned above, the phenomenon is an example of three-wave mixing; in this case two components of the input laser beam make up the two factor waves, and the BEAM SPLITTING, amplification and frequency tripling of the laser are carried out in the large room shown. The bright areas on the far wall (160 feet from the camera) are caused by flash lamps that energize the laser amplifiers. The laser beams are infra­ red, and so they are invisible. They are converted into invisible ul­ traviolet light by crystals in the six boxlike modules. The visible OMEGA 76 higher-harmonic output is the product wave. The two components of the in­ put beam arise as the beam enters the crystal because of asymmetries in the alignment of the atoms that make up the crystal lattice. When the oscillating electric field of the laser enters the crystal, it displaces electrons from their equilibrium loca­ tions, just as it does in the corona of the fuel pellet. In a KDP crystal the electrons are harder to displace along a preferred axis called the optic axis, or z axis, than they are along any direc­ tion lying in the x-y plane perpendicu­ lar to the z axis. The crystal is "stiffer" to the force of the electric field along the optic axis, and the electromagnetic wave whose electric field is aligned with the optic axis propagates the fast­ est through the crystal. Imagine that the crystal is cut so that its face makes an oblique angle to the z axis, and suppose a light ray strikes the crystal at a right angle to its face [see illustration on page 72]. The electric field of the ray, which then oscillates parallel to the face of the crystal, has two mutually perpendicular compo­ nents. The first component, which lies light emerging from some of the modules comes from heat lamps, switched on intermittently to keep the crystals at a constant tem­ perature. Residual green light from the conversion causes the green glow. The beams emerge from the modules in groups of four, and their energy is measured. Subsequently the beams are reflected by the mirrors visible in the foreground into the adjacent target area. © 1986 SCIENTIFIC AMERICAN, INC in the x-y plane and is parallel to one of the crystal axes (by convention the y axis), is called the ordinary, or 0, ray. Because it has no component in the z direction, it propagates relatively slowly through the crystal. The second component of the incident ray is per­ pendicular to the first one in the plane of the crystal face. This second com­ ponent is called the extraordinary, or e, ray, and it does have a component in the z direction. Because of its z com­ ponent, which depends on the orienta­ tion of the crystal face with respect to the z axis, the e component of the inci­ dent beam moves through the crystal faster than the 0 component. When high-intensity laser light trav­ els through a crystal, the electrons dis­ placed by the electric wave tend to feel a restoring force that is neither in the same direction as their displacement nor proportional to it. It is the nonlin­ ear, or nonproportional, response of the electrons to the e and 0 compo­ nents of the electric field that gives rise to a so-called nonlinear current wave. The current wave is made up of elec­ trons oscillating in the x-z plane at twice the frequency of the input laser beam, and it moves through the crystal at a velocity equal to the average ve­ locity of the e and 0 components of the input laser. More precisely, the refrac­ tive index of the current wave is the av­ erage of the refractive indexes of the e and 0 components. The current wave of oscillating elec­ trons generates a wave of laser light at the second harmonic of the laser beam, just as oscillating electrons in an antenna generate radio waves. The second-harmonic light wave can be generated efficiently, however, only if it propagates at the same velocity as the current wave: it must "surf," or be in phase with, the current wave. Be­ cause the second-harmonic wave is an e wave, its velocity and therefore its re­ fractive index vary with the angle be­ tween its direction of propagation and the optic axis. Hence the velocity of the second harmonic can be matched with the velocity of the current wave by cutting the crystal in such a way that the direction of propagation of the laser through the crystal makes a predetermined angle with the optic axis [see illustration on page 73]. Attaining High Compression Two more features of laser-fusion design, complementary to the require­ ment for a short-wavelength input beam, are crucial for reaching high fuel compressions in the pellet. They are the uniformity of the illumination and absorption of the incident laser 10" '10" 0 ....l W 1012 >= z 0 10" a: I:::J w z 10" •. 35 •. 5 10' 10· MICRON MICRON .1 MICRON 10' 10' 103 10' LASER ENERGY (JOULES) NEUTRON YIELD obtained with several laser systems depend� primarily on laser wave­ length, energy and the symmetry of the incident beams. The sloping lines on the graph indicate the energy gain of the pellet: the fusion energy generated divided by the input laser energy. Results are from KMS Fusion, Inc., of Ann Arbor, Mich. (KMS); Lawrence Liver­ more National Laboratory (LLNL); the Laboratory for Laser Energetics at the University of Rochester (LLE), and the Institute for Laser Engineering in Osaka, Japan (OSAKA). beams and the hydrodynamic stability of the imploding pellet. The uniformity of the illumination and absorption is important for purely geometric reasons. If the radius of the compressed shell is to be decreased by a factor of 30, the implosion velocity must be uniform over the surface of the shell to within about one part in 60. That requirement limits the allowable fluctuations of the laser irradiation in­ tensity to one or two parts in 100 over the surface of the pellet. How many beams are needed, and how should they be arranged around the pellet? For laser systems employ­ ing four, six, eight, 12 or 20 beams, each beam can be positioned as if it were at the center of a face of one of the five Platonic solids (tetrahedron, cube, octahedron, dodecahedron or icosahedron) and directed perpendicu­ lar to the face. The beams are then dis­ posed symmetrically around the pellet if it is placed at the center of the Pla­ tonic solid. The frequency-doubled VULCAN laser at the Rutherford Apple­ ton Laboratory in England, and the GEKKO-XII laser in Osaka both employ a dodecahedral geometry. When the number of beams is greater than 20, or in other words greater than the num­ ber of faces in the regular icosahedron, suitable geometric configurations that maximize the uniformity of the illu­ mination can be calculated. For exam­ ple, a 32-beam system can be based on the configuration of the 32 pentagonal and hexagonal faces of a soccer ball. Symmetric placement of the sur© 1986 SCIENTIFIC AMERICAN, INC rounding beams does not guarantee uniform irradiation of the pellet, how­ ever: not all points on a hemisphere of the 'pellet are equally irradiated by a given beam. The attainable irradia­ tion uniformity increases as the num­ ber of beams is increased, although the complexity of the laser system also increases. The number of beams is therefore a trade-off between uni­ formity and complexity. The 32-beam sys-tem appears to be an attractive compromise. The optical quality of each beam and the balance of energy among the beams are also important to the uni­ formity of irradiation. In our labora­ tory we have put much emphasis on generating laser beams with a uniform cross section, and we have been able to reduce the intensity variations across the profile of a beam to less than 10 percent. Other methods of enhancing the illumination uniformity have been proposed by workers in Osaka. and at the Naval Research Laboratory in Washington. In both methods optical systems near the focusing lens frag­ ment each beam that is split from the master beam into many-say 400smaller beamlets. Each beamlet is giv­ en a random phase, and each spreads its energy over a full hemisphere of the pellet. Any spatial nonuniformities in any of the beams at the focusing lens are then averaged out over the pellet surface. Initial results from both labo­ ratories show that the methods im­ prove the uniformity of irradiation. Another major goal in the study of 77 laser fusion is to understand and con­ ium fuel. At the same time the fuel trol hydrodynamic instabilities of the can rise through the dense plasma in fuel pellet. When the imploding shell is bubbles. The instability is similar to a decelerating just before the peak com- . phenomenon of incompressible fluids pression of the fuel, the dense fluid called the Rayleigh-Taylor instability, of the plasma shell can fall in spikes in which a dense fluid falls through a and mix with the deuterium and trit- light one. It can be triggered by non- uniformities of irradiation or by im­ perfections in the shape or composi­ tion of the pellet, particularly when such imperfections are small. The ab­ lation surface outside the fuel lay­ er can also become hydrodynamically unstable as the implosion accelerates. [jJ z o a: u � w U z � (f) 15 100 � 100 ���__����__L--L�L-�����L-�����__L-����__��������__���__� __ 80 60 40 o 20 20 40 60 80 100 DISTANCE (MICRONS) COMPUTER SIMULATION of the implosion of a fuel pellet de­ signed for a reactor illustrates the effect of irradiation nonuniformi­ ties on pellet performance. The nonuniformities are assumed to be symmetric for rotations about the north-south axis of the pellet. They are characterized by the number of maximums of irradiation intensity encountered around the great circle represented by a line of longitude on the pellet. The pellet is a plastic shell with a layer of cold, liquid deuterium and tritium fuel on its inner surface and a small amount of deuterium and tritium gas inside. When the gas is compressed, it forms a so-called spark. plug region in which thermo­ nuclear reactions are ignited. An irradiation nonuniformity of 1 percent is imposed in the simulation. The illustration shows the 78 compressed core just before ignition; darker regions correspond to higher densities. The grid lines indicate the resolution with which the evolution of the pellet implosion was calculated. The boundary between the main fuel and the gas, which was the inner layer of the shell before the compression, is shown as a white contour line. Most of the material shown is compressed to a density of between 100 and 500 grams per cubic centimeter, or from 500 to 2,500 times its liquid density. The calculation was done with the program ORCHID, written by Charles P. Verdon of the University of Rochester; it required tens of hours on a CRA Y X-MP, one of the world's fastest supercomputers. In spite of the nonuniform compression the fusion yield in the simulation was nearly 100 times the laser-energy input. © 1986 SCIENTIFIC AMERICAN, INC That instability can lead to a breakup of the shell, a mixing of the fuel and the shell and a smaller peak compres­ sion. Ultimately such effects eliminate or reduce the thermonuclear yield. One-dimensional spherically sym­ metric simulations show that the high­ est compressions can be obtained for pellets whose shells are thin compared with their radii. Hydrodynamic insta­ bility, however, is potentially most dangerous for pellets with thin shells. A realistic pellet design must therefore represent a trade-off between the de­ sirability of relatively thin shells and the constraints imposed by uniformity and stability considerations. It seems desirable to ensure that the shell nev­ er becomes thinner than a few percent of the radius of the pellet throughout the implosion. Two designs currently assume ma­ jor importance. The shell in our simu­ lation is made out of plastic, and it contains an inner, cryogenic layer of liquefied deuterium and tritium. The shell interior is filled with low-density deuterium and tritium gas at the vapor pressure of the cryogenic liquid. The use of plastics made out of elements of low atomic number reduces the emis­ sion of radiation from the plasma co­ rona, which could preheat the fuel. Since the cryogenic pellet is essentially empty, the shell is free to accelerate in­ ward until the final moments when its kinetic energy is converted into heat. An important alternate pellet de­ sign, which we can mention only brief­ ly, exploits the response of materials of high atomic number to laser irradia­ tion. The fuel pellet is placed inside a so-called radiation case made out of a material such as gold. When the interi­ or of the gold case is irradiated by the laser, a large fraction of the incident energy is transformed into X rays. The X rays then irradiate the pellet and cause it to implode in a highly uniform way. Since the laser beams do not irra­ diate the pellet directly as they do in the first design, this alternate scheme is called indirect drive. Computer Simulations We shall conclude by describing computer simulations of a fuel pellet that might be considered for a com­ mercial reactor, irradiated by a kryp­ ton-fluoride laser whose energy is 1.6 million joules and whose wavelength is a quarter of a micron. (A joule is the energy needed to lift one kilogram approximately 10 centimeters.) In our simulations the laser energy is deposit­ ed with an average power of 3 X 101 4 watts i n a pulse whose intensity in­ creases steadily for five to 10 nanosec­ onds. The intensity of the pulse is kept low enough to avoid generating signif­ icant numbers of suprathermal elec­ trons in the plasma corona. One important measure of the re­ sponse of the pellet to the laser is called the spark-plug convergence ra­ tio, the ratio of the initial radius of the spark-plug material to its final radius. To maintain the stability of the pellet the ratio should not be too high; we aim for a ratio of about 50. In our models we then impose nonuniformi­ ties from point to point in the laser il­ lumination of the pellet. As the inter­ face between the shell and the fuel decelerates near the time of peak compression, pressure variations and the growth of hydrodynamic instabil­ ities cause the density contours to fol­ Iow a complex pattern around the core [see illustration on oppOsite page]. The alpha particles emitted by ther­ monuclear burning in the spark-plug region begin to deposit their energy in the main fuel layer. When the prod­ uct of the density and the radius of the spark-plug region rises to values large enough to stop most of the al­ pha particles, thermonuclear igni­ tion is achieved. The temperature in the spark-plug region reaches approx­ imately 100 million degrees C.; the temperature of the main fuel layer is about 30 million degrees. The ther­ monuclear burn propagates radially outward into the main fuel layer, the temperature profile becomes smooth­ er and the burn becomes nearly spheri­ cal as the main fuel is consumed. The pellet then disassembles and the ther­ monuclear fire goes out. According to our simulations, which have considered a variety of irradia­ tion nonuniformities, the thermonu­ clear energy output is typically 100 times the energy input of the laser. These results inspire renewed confi­ dence in the feasibility of laser fusion. A power plant capable of generating almost a billion watts might be feasi­ ble if 10 such fuel pellets were ignited every second and if the overall effi­ ciency of the laser were approximately 15 percent. On the basis of these calcu­ lations we expect that a laser emitting more than 1.6 million joules of energy but less than 10 million joules would be appropriate for a reactor. In summary, considerable progress has been made in understanding the physics of laser fusion, and the need for short wavelengths is now firmly es­ tablished. The results of the next dec­ ade of experimentation should resolve the basic scientific question for laser fusion: Can ignition be achieved? An affirmative answer would bring the promise of abundant fusion energy for the 2 1st century one important step closer to its practical realization. © 1986 SCIENTIFIC AMERICAN, INC "Cameron has never heard the love in his parents'voices!' �ct � He's deaf. Cameron Garberoglio is one of 16 m i l l ion hearing-impaired Americans who need your help. And ours. The Deafness Research Foun­ dation is the only national volun­ tary health organization solely committed to finding the an­ swers heari ng-impa i red Amer­ icans a re wa iting for. The Foundation is unique-its overhead is funded by more tha n 2,000 doctors, a l lied pro­ fessionals, and medical societies. So 100% of your contri bution goes d irectly i nto research. Research that might one day help Cameron hear his parents, and the crack of Regg ie's bat. Take a moment, and send your tax-deductible contribution to the Deafness Research Foun­ dation today. Because there's so much to hear. Help him. There's so much to hear. Deafness Research Foundation New P.O. Box 5000 New York 10017 79