Many life-cycle analyses have been made of both silicon and thin-film solar cells. In 2007, Raugei et al. [10] published a careful study of the environmental effects of both silicon and thin-film solar cells using actual production data from Europe. For polycrystalline silicon (the most common kind), one had to decide where it came from. If it came from the electronics industry, even if it is the off-grade rejected material, the energy cost is very high, as shown in Box 3.5. On the other hand, if the solar industry grows to the extent that it can build its own factories to produce solar-grade silicon of lower purity, the energy cost would be much lower. Both the worst-case and best-case scenarios for silicon were compared with CdSe and CIS (copper indium diselenide) thin-film systems. (CIS is similar to the CIGS men­tioned above.) The results for energy payback time are shown in Fig. 3.42.

Here is what went into these calculations. First, the materials used were listed. For thin film, these were glass, plastic, water, and the electronic layers. Glass was

by far the largest part, and the thin films the smallest. Then the energy to make these materials and the electricity to fabricate the cells in the factory were evaluated. That is for just the bare cell. To this must be added the balance-of-system; namely, the parts needed to complete functioning modules and arrays. These include aluminum for the frame, steel for the supports, cables and connectors, and the electric equip­ment for converting DC to AC at the grid voltage. There is also fuel oil used in installation. The energy cost of decommissioning and recovery of materials at the end of life was not included, but these were considered by Fthenakis [11]. The energy used was assumed to be the mix of fossil and hydro energy typical of Europe, with 32% average efficiency in generating electricity. As for the solar energy output, the assumptions were quite conservative. The sunlight available was 1,700 kWh/m2 per year, typical of southern Europe, not a desert. A 25% efficiency loss was assumed to account for dust accumulation and electrical equipment. The lifetime of the system was taken to be only 20 years.

The calculated energy payback times are shown in Fig. 3.42. As expected, it is very long for silicon in the worst case, when it is obtained from the electronics industry. However, if special factories are built to produce solar-grade silicon in ribbon form, the payback time is competitive with thin film. CdTe is the clear win­ner in this study, its payback time being only 1.5 years. The graph also shows the breakdown between the energy costs of the bare cells and the balance-of-system or BOS. Note that for CdTe cells, it is the BOS that takes the most energy to make. The global warming potential (CO2 emissions) of these systems is usually also calculated in these studies, and of course it is much smaller than that of fossil-fuel energy sources. After initial greenhouse gas (GHG) emissions during buildup, a solar plant produces electricity with almost no emissions for up to 30 years.

Cadmium is a very toxic element. In 2009, there was an uproar because some toys imported from China were found to contain cadmium in ingestible form. However, that does not mean that a compound like CdTe is toxic. Salt, NaCl, is certainly not dangerous although sodium and chlorine are themselves very toxic elements. In the case of CdTe, one worries that Cd could be emitted into the envi­ronment during manufacture and operation, even though the cells are encapsulated in glass and Cd is very stable, with almost zero evaporation. Unknown to most people, incidental emission of Cd also occurs in coal and oil plants. Raugei et al. [10] estimated the emission of Cd from a solar plant and found that it is 230 times smaller than from a coal plant for the same energy output! Detailed evaluations of dangers from toxic substances have been done by Fthenakis et al. [12, 13].

The amount of land used in solar power and the environmental impact on it has been compared with other energy sources by Fthenakis and Kim [14]. Not surpris­ingly, these solar proponents find that solar energy requires the least amount of land and biomass energy requires the most. The use of land in coal and nuclear power includes the land destroyed in mining and waste storage. Hydroelectricity uses dams which convert land into lakes. However, the usage of the area may actually be improved, and wildlife may only be changed from animals to fish. A large area covered with solar arrays may still allow desert animals, birds, and tortoises to live if some plants are allowed to grow under the panels. However, the reflectivity (albedo) of the desert will be decreased by the absorbing black panels. A very large area of these may affect cloud formation and the entire climate in the region.

The life-cycle studies of solar power are less complete than those of wind power and seem to be optimistic. The wind studies included replacement of parts as they wore out and the energy costs of inspection and maintenance, including the gasoline usage by the inspectors. Dust will cover the solar panels and should be washed off. In the desert, there is no water. The glass covers of the panels will be blasted in sandstorms. In temperate climates, plants will grow and have to be pulled out before they get too high. With no weeding, a solar farm will be immersed in a dense forest in 10-20 years. Space must be left open between rows of panels for machines to do this. The Mars rovers have experienced what happens to solar panels without main­tenance. Dust accumulated on them, decreasing their power. The Rovers depended on wind storms to blow the dust off. After years of dust accumulation, the power became so weak that communication became difficult. The rovers had to be manipu­lated onto a crater’s edge to tip the panels to face the sun more directly. In solar farms on earth, the panels are fixed.44 It has been estimated that mechanisms to track the sun would add 25% to the cost of the panels but could increase their capacity by 40%. The cost of energy storage for night time was not included in these studies.

However, the storage problem was addressed in admirable detail by Mason et al. [15]. The only method being considered is CAES, which is described in the Wind Energy section (Fig. 3.17). The electrical energy being stored is used to compress air into these caverns. When the energy is needed, this compressed air is released and used to help drive a gas turbine to produce electricity. CAES has been tested only at two places: in Germany, where a 290-MW plant has been operating since 1978, and in Alabama, where a 110-MW plant has been operating since 1991. These CAES systems were used to store excess electricity produced conventionally in off-peak hours. There are numerous sites in the USA where caverns suitable for CAES exist, but they cannot be close to the solar farms for several reasons. The deserts where there is the most sunlight have few suitable sites and insufficient water needed for cooling. They are also far from population centers. A system of high-voltage DC (HVDC) transmission lines is proposed to connect the solar plant to the storage plant. The energy capacities of the two plants also have to be matched.

The Mason study [15] considered a storage plant that provides peak power 10 hours a day, Monday through Friday when it is needed, and another for base­load power 24 hours a day for a future central-station solar farm. The daily solar output during the year was calculated, as well as the storage requirements during each day. The costs of the solar and storage plants were carefully itemized, includ­ing such items as maintenance, land preparation, interest during construction, and replacement of parts. The HVDC cost was included, as well as the substations for converting DC to AC. The results for a peak-load PV-CAES system are summa­rized in Fig. 3.43. The cost of electricity from PV systems with storage is compared with that from an advanced-cycle natural gas plant with carbon sequestration. In the next 10 years, the cost of the PV part is expected to go down, but the CAES part does not go down as much. It accounts for a third of the total cost. Solar electricity for peak loads, it appears, will be competitive with that from natural gas by 2020.

For base loads, however, PV-CAES electricity will cost $0.118/kWh, considerably more than the $0.076 and $0.087, respectively, from gas and coal plants, both with CCS. This is all conjectural, however, since the cost, safety, stability, and legal problems of underground storage have never been tested on a large scale.

The motion of tides, currents, and ocean waves can be used to produce energy. A few places, like the Bay of Fundy, have high tides, and the water rushes through a narrow channel four times a day. If the speed is greater than about 5 knots (2.6 m/s), the current can drive an electric generator, but there are very few such sites. A new method called Vivace87 is claimed to work at speeds as low as 2 knots (1 m/s). Flexible cylinders are anchored to plates on the sea bottom. Currents flowing back and forth make the cylinders flex and wobble, and this motion is used to generate electricity. How they do that and what the cylinders are made of are not revealed. Tides and waves also make the ocean level go up and down. Several systems using this effect are based on the same principle. A rigid tube is anchored to the sea bot­tom. A diaphragm inside the tube is driven up and down by a buoy floating on the top. As the buoy moves up and down, the diaphragm drives air in and out of an aperture at the top end of the tube. This flow of air turns a turbine to generate elec­tricity. An underwater cable carries the electricity to shore. This method requires a floating object that can be seen and collided into.

The most publicized system is the Polamis (“sea snake” in Greek),88 designed to capture wave energy. It looks like a series of giant sea snakes floating in the ocean. Each snake points in the direction of wave motion, perpendicular to the wave crests, and consists of metal cylinders with the size of railway cars hinged to each other so that the snake flexes with the waves. In between the cylinders are air pistons push­ing air back and forth with the wave motion. This air drives onboard electric gen­erators. These Polames have been built in several countries, Portugal for one.

Cost and power have been calculated, but none of these ideas has been worked out for impact on the environment, wildlife, and ship traffic. Engineering for 30-year lifetime in the sea may be difficult. The resistance of materials against salt water dam­age, so important in offshore wind turbines, has not been mentioned, for instance. The power is also not constant, so that some storage mechanism is needed to level it out. It is clear that these entrepreneurial ventures cannot yet be taken seriously.

The reason that a toroidal magnetic bottle has to have twisted field lines is that ions and electrons drift vertically in opposite directions, as explained in Chap. 4. This drift arises from the fact that the magnetic field is necessarily weaker on the outside of the torus than on the inside, near the hole in the doughnut. An obvious idea to get a larger volume of plasma without changing the drifts is simply to make the tokamak taller, without changing its radius. This is shown in Fig. 7.11a. The sharp corners have very bad curvature, so they have to be rounded off. A machine built at General Atomics in San Diego, the Doublet, is shown in Fig. 7.11b. This looks like two merged tokamaks, one on top of the other, connected by a region with good curvature. The bean-shaped cross section studied at Princeton University has good curvature on the inside of the torus and shown in Fig. 7.11c [4]. It turns out that it is not necessary to curve the inside surface; keeping it straight is almost as

good, since the magnetic field naturally gets stronger as the plasma tries to escape toward the inside of the doughnut. We then have the D-shape shown in Fig. 7.11d. The outside of the D still has bad curvature, but it curves more gently than in a circular tokamak because of the elongation. Figure 7.12 is a D-shaped toroidal-field coil shown during the construction of the ASDEX tokamak in Germany. This was one of the first large tokamaks of the time (ca. 1980) but is small compared with those operating today.

The D-shape is not all gravy: the bad curvature at the corners of the D is very sharp, but at least it occurs in only a small part of the total surface. Actually, this part of the D can be used for a necessary function — that of plasma exhaust. A product of D-T fusion is helium (alpha particles). This “ash” has to be taken out since confining it would use up the magnetic confinement capability reserved for the DT. Furthermore, the normal escape of DT plasma, though slow, still carries out

Fig. 7.12 A D-shaped ASDEX coil more heat than the walls of the chamber can stand. By channeling the escaping plasma into the corners of the D, special devices called divertors can be placed there to handle the heavy heat load. Figure 7.13 shows a diagram of the cross section of a D-shaped tokamak with divertors. The last closed magnetic surface is changed with locally placed coils so that the field lines leave the surface and lead outwards into the divertor. Plasma diffusing to that surface then enters the divertor, where it is captured by high-temperature, rapidly cooled materials.

Figure 9.21 shows what a niobium-tin cable looks like inside. There are over 1,000 strands in six bundles. At the center is a helix making room for the pipe that carries the liquid helium. The outer casing is a stainless steel jacket 37.5 mm (1.5 in.) in diameter. This cable, designed for the toroidal field coils of ITER, can carry 80 kA at 9.7 T. Each strand is about 0.8 mm in diameter and consists of a Nb3Sn filament sheathed with chromium and covered with about as much copper as Nb3Sn. The copper is necessary to mitigate quenches. A quench occurs when part of the superconductor goes normal, losing its superconductivity because of over­heating or over-current. Huge voltages would build up as the current tries to force

its way through a normal conductor with resistance, and there could be an explosion. Copper can make this a gentler accident. The complexity of superconducting cables is bad enough, but to wind them into magnetic coils means that each cable has to be over 1.5 km (a mile) long.

A tokamak has many different kinds of magnet coils, and each requires a different design. Some of these can be seen in Fig. 9.22. The toroidal field (TF) coils are the large D-shaped coils. They operate up to 6 T and are the heaviest ones. Transporting them to the ITER site requires special barges, trucks, and roads. The large, horizon­tal ones encircling the machine are the poloidal field (PF) coils, which give the field lines their twist and shape the plasma. Because of their size, they cannot be trans­ported and must be wound on site. The coil winding building at the ITER will be 253 m long, 46 m wide, and 19 m high.3 A critical component is the central solenoid (CS), seen inside the hole in the torus. There is very little space there, and most of it is taken up by the interior blanket modules. This coil is the other half of the PF system that shapes the plasma and drives the tokamak current. The CS is 13 m tall and 4.3 m in diameter, weighing 1,000 tons. It also produces the highest field of 13.5 T. Figure 9.23 shows a test section of it that has been made.

There are smaller coils besides these main coils, but the difficult part is to join the superconductors to their feeds. Current is fed into the coils from normal­conducting cables, and then a superconducting switch is turned on so that the

current flows only in the superconductors and the feed cables can be disconnected. These junctions are very complicated, especially since the current has to go through the wall of the cryostat from room temperature to 4 K. Almost all the nations supporting ITER participate in designing and producing the magnet system. Some make the NbTi and Nb3Sn materials. Some make it into strands and cables. Some wind the cables into coils. And some make the feed cables and the junctions. The technology has already been developed for smaller tokamaks, and the steps to ITER, DEMO, and reactor are only matters of scale.

Mirror machines, together with stellarators, were the strongest proposals for plasma confinement devices when fusion research started in the early 1950s. The effort was led by R. F. (Dick) Post, who is still actively pursuing the mirror concept today. Unfortunately, the mirror program at Livermore was canceled in 1986 by the USA in order to concentrate on the tokamak. Research on mirror confinement has con­tinued in Russia, under the guidance of Dmitri Ryutov, and in Japan, with the Gamma 10 machine. Reactors using the mirror principle would have the great advantage of direct conversion of energy to electricity without a thermal cycle, the same advantage that hydro, wind, and solar power have over other power sources.

A mirror machine is a leaky magnetic bottle. In Fig. 10.20, a pair of coils gener­ates a magnetic field that bulges out between them. An ion or electron will gyrate in a Larmor orbit around the field lines, as shown in Fig. 4.10. As the orbit approaches the strong-field region at the ends, the Larmor orbit becomes smaller and smaller. To conserve angular momentum, the particle has to gyrate faster and faster, just as a skater does when she pulls her arms in during a spin. But energy is conserved, and to get this extra rotational energy, the particle has to take it out of its translational motion. It slows down in its efforts to escape out the end. Finally, all its translational energy is lost, and the particle has to turn around and go back.

It is reflected by the magnetic mirror, which is just the strong-field region. The particle bounces back and forth between the mirrors at each end.

When a plasma is created, ions (and electrons) have both translational energy and gyrational energy. The ones with lots of translational energy and little gyra — tional energy are lost out the ends. If the mirrors are strong, meaning that the mirror ratio between the fields at the throat and at the midplane is large, only a few particles are lost; and the rest of the plasma is confined. However, this plasma is not in thermal equilibrium; some of the velocities are missing. These velocities are in the loss cone. The plasma will then devise an instability to regenerate those missing veloci­ties and fill the loss cone. There is then a continuous loss, giving magnetic mirrors a short confinement time. Such microinstabilities, however, are not the main prob­lem in mirrors.

The gasoline engine is a marvelous piece of engineering. Honed over hundreds of generations of models, it fires an explosion thousands of times a second, and yet we can hardly hear it as it smoothly pushes the car through the air. What is wrong with it? It uses gasoline very inefficiently, and it emits carbon at a rate equivalent to throwing a charcoal briquette out the window every quarter mile.

Electric cars are even quieter… so silent that it has been proposed to put a noise generator in them to warn pedestrians. Electric cars have no emissions, but they get their electricity from power plants that emit GHGs. However, power plants burn fossil fuels much more efficiently than cars do, so the total emissions are lower. It is because power plants run at much higher temperatures than cars can, and the Carnot efficiency (see Chap. 2) is much higher. There is a big difference between 40 and 15% efficiency, and most people do not realize this. The main problem with electric cars is the battery. There is no type of battery of reasonable size and weight that can take a car 300 miles on one charge, and it takes many hours to recharge the battery. If you run out of “gas” in an electric car, you would have to stay in a motel with a plug. But electric cars have great advantages. We will consider these next and hybrids later.

Efficiencies of Gas and Electric Cars

A normal car can use only about 15% of the energy in gasoline, though some say it could be 30%. The breakdown is shown in Fig. 3.52. Most of the energy is lost in heat, 30% in the radiator and 30% in the exhaust from the muffler. A few percent more is lost in the engine and in the transmission line between the motor and the wheels. Fully 17% is used in idling while the car is not moving, such as at a red light. The motor has to be running so that it can start again rapidly. Accessories such as lights and radio take only 2%. That leaves only 12.6% for propulsion of the car.

PROPULSION

12.6%

About half of this is lost as heat in the brakes to stop the car. The rest, 6.8%, is all there is left to move the car!

Electric cars store energy in a battery bank and use that to drive a motor that drives the wheels. The battery may get a little warm, but the heat energy lost is trivial compared with the 60% in normal cars. The stand-by energy is saved since the motor simply turns off when the car is coasting or stopped. The braking energy is recovered into the battery, though the brakes will get a little hot, and that energy will be lost. The accessories, including the lights, the radio, and the computer, will take a few percent, and so will the transmission, but all the rest is available to move the car. Electric cars can convert about 75% of the energy stored its battery into useful power. The battery is charged with electricity from the grid, and the environmental impact of that process depends on the location. In most places, coal or natural gas is used to generate electricity, and GHGs are emitted. However, this is better than burning oil products in cars for several reasons. Power plants can be 40% efficient, three times better than cars. So less fuel is consumed and less CO2 is emitted. Furthermore, power plants can be located some distance from cities, thus sparing them from pollution. Electric vehicles emit only water. In locations where hydroelectric or nuclear power is available, the air is even cleaner. Even noise pollution is abated.

Vehicles running totally on electricity are being used successfully in service vehicles and golf carts, which do not have far to travel. The Tesla Roadster has shown that electric cars can have sports-car performance at a price. The big buga­boo is transportable energy. There is no known type of battery that will carry a car 300 miles and recharge in 5 min, as we can get from gasoline. Meanwhile, we can save on gasoline by using hybrids. These will be discussed next, followed by battery prospects.

If we had straight field lines in a cylinder, there would be no problem; but when we bend the cylinder into a torus so that the field lines do not strike a wall, the first of several toroidal effects comes into play. In Fig. 4.8b, we saw that an electromagnet generates a magnetic field by driving electric current in a coiled wire. The field lines are then formed inside the wire coil. When we bend the cylinder into a torus, as in Fig. 4.13a, the wire coil will also have to be bent to surround the torus, resulting in the configuration shown in Fig. 4.14. Each turn of the coil carries current in the direction of the arrows, generating a magnetic field purely in the toroidal direction.

Two of the field lines have been drawn. Notice how the coils crowd together as they go through the hole in the torus. The bunched current then creates a larger field at point A than at point B, which is farther from the doughnut hole. The magnetic field is always larger on the inside of a torus than on the outside. This is a toroidal effect that does not happen in a cylinder. The consequence of this effect is that charged particles no longer gyrate in perfect circles. Let’s look at the orbit of an ion in the right-hand cross section of the torus in Fig. 4.15. Normally, it will gyrate clockwise in a circular orbit, but here its orbit has been distorted into a spiral. Remember that it is the Lorentz force that makes the ion gyrate, and this force is proportional to the magnetic field strength. On the left-hand side of the orbit, the ion will feel a stronger force than it does on the right-hand side, where the field is weaker, so it will turn more tightly on the inside. The result is that the ion’s guiding center drifts downwards in this diagram. Observe that an electron drifts upwards because it has negative charge, and therefore gyrates in the opposite sense to that of the ion. This drift has been greatly exaggerated here, but nonetheless it has a huge effect on the plasma, collecting the positive charges on the bottom and the

negative charges on the top, as shown. These charge bunches will create an electric field going from the positive charge bunch at the bottom to the negative one at the top. Such a vertical electric field, as we shall see, will blow the whole plasma out toward the outer wall. Such a simple magnetic bottle will not work!

This problem was recognized very early in the game. The famous astronomer Lyman Spitzer, Jr., was riding up the long ski lift at Garmisch-Partenkirchen when he thought of a solution. Incidentally, Spitzer was the prime mover in getting the Hubble telescope built. It was not named after him, but after his death a Spitzer telescope was finally put into orbit. Spitzer’s solution was to twist the torus into a pretzel shape, as in Fig. 4.16. If you were a particle traveling along the depicted field line starting at B, you would feel a stronger magnetic field on the left than on the right. When you reach A, the strong field is on the right, now that the torus has been twisted. This is different from the circular torus of Fig. 4.14, where the strong field is always on the same side. Let’s look at the two cross sections in Fig. 4.16 in more detail. These are shown larger in Fig. 4.17. Cross section A is the same as that in Fig. 4.15, with the magnetic field pointing out of the page and with the ions drifting downwards. In cross section B, on the opposite side, the field also points out of the page instead of into the page, as in a circular torus. The fat arrows are

supposed to show this. With the field in the same direction, the ion gyrates clockwise in B, as it does in A. In B, however, the strong field is on the right side of the orbit, so the ions drift upwards instead of downwards. The vertical drifts then cancel as the ion moves along the figure-8 along a field line, and the cata­strophic separation of charges, in principle, does not occur.

This type of magnetic bottle was named a stellarator by Spitzer because it was intended to reproduce the conditions inside stars which allow them to generate fusion energy. A series of a half-dozen figure-8 stellarators was built at the Plasma Physics Laboratory in Princeton University in the 1950s to test this confinement idea. A model of a figure-8 stellarator (Fig. 4.18) was shown at the 1958 Atoms for-Peace conference in Geneva, in which thermonuclear fusion was declassified and different nations showed off their inventions. The individual coils carrying the current to generate the magnetic field can clearly be seen in this model. There was also an electron gun inside the chamber that could emit electrons that visibly traced the magnetic field lines. In addition to this model, an entire real, working stellarator was shipped to Geneva and reassembled there in the US exhibit. The Russians proudly displayed their Sputnik satellite, but their fusion exhibit was an unim­pressive, unintelligible black hunk of iron called a tokamak. It was only many years later that the world realized that that was the real star of the show.

What Have We Accomplished?

A controlled fusion reaction requires holding together for a long enough time a plasma that is hot enough and dense enough. These critical conditions can be quan­tified by the triple product Tut, a modification of the Lawson criterion explained in Chap. 5. Here, Tis the temperature of the ions, the reacting species; и is the density of either the ions or the electrons, since the plasma is quasineutral; and t (tau) is the energy confinement time, a measure of how fast (or slowly) energy must be applied to keep T constant. Over the years, over 200 tokamaks have been built, and the value of Tut achieved in each has been calculated. Some of these are plotted in Fig. 8.1 as a function of time. This measure of success has increased over 100,000 times in four decades, recently doubling every two years.

Most of this increase has come from the confinement time. The first experimental machines suffered from hydromagnetic instabilities such as the Rayleigh-Taylor and the kink instabilities described in Chap. 5. These can take the plasma to the wall at the speed of a field line wiggle called an “Alfven wave,” which limits the confine­ment time t to microseconds. Once these were controlled, t increased a thousand-fold to several milliseconds, at which point microinstabilities were the limiting factor. After years of understanding banana orbits, magnetic islands, ballooning modes, and connection lengths, these instabilities were minimized; and t increased another thousand times to the present value of several seconds.

The rate of progress in fusion can be compared with that in the development of computer chips, the famous Moore’s Law. Gordon Moore had predicted that the number of transistors on a chip would double every two years, an unbeliev­able rate which was actually followed almost exactly. Figure 8.2 shows how this growth compares with a range of doubling times. The fusion figure of merit in Fig. 8.1 keeps pace with Moore’s law, now also doubling every two years. Both of these outstrip Livingston’s law for particle accelerators; where the energy doubling time is three years.

‘Numbers in superscripts indicate Notes and square brackets [] indicate References at the end of this chapter.

F. F. Chen, An Indispensable Truth: How Fusion Power Can Save the Planet,

DOI 10.1007/978-1-4419-7820-2_8, © Springer Science+Business Media, LLC 2011

Fig. 8.2 Moore’s Law for semiconductors compared with doubling rates

Here are pictures of the four large tokamaks which provided the points at the top of these graphs (Figs. 8.3—8.6).1

As you can see, or cannot see, the tokamak itself is hidden behind a jumble of equipment which includes the neutral-beam injectors, power feeds to the coils, the support structure, and diagnostic instrumentation. To show the size of these machines, Fig. 8.7 is an inside view of the vacuum chamber of DIII-D when it is opened up to air.

In a burning plasma, 3.5-MeV alpha particles are generated, and as they cool down they transfer their energy to the plasma, keeping it hot. Before they become thermalized, however, the alphas are in the form of beams streaming along the magnetic field lines, and beams can excite instabilities. To do this, the velocity of the beam has to coincide with the velocity of a wave in the plasma; and the syn­chronism causes the beam energy to be transferred to the wave. The wave can become

so strong that it disrupts the plasma. There is a plasma wave called the Alfven wave that travels along the B-field and can have just the right velocity to match that of the alpha-particle beam. The danger that this can happen can be predicted precisely by theory [19], but whether it will actually happen or not depends on the details. ITER will be the first machine that can test for Alfven wave instabilities in a D-T plasma. If these turn out to be important, their avoidance is a physics problem that needs to be solved.

The beauty of inertial confinement was supposed to be its freedom from the instabili­ties in magnetic confinement. No such luck: there are new instabilities! First there is an old one, the Rayleigh-Taylor instability (Chap. 5), which occurs whenever a light

fluid pushes against a heavy one. The expanding plasma pushes against the capsule with a huge force. If there is any deviation from smoothness in either the capsule or the laser light, small ripples will grow and destroy the compression before it gets very far. Figure 10.40 shows what can happen.

Parametric instabilities are a new class of instabilities caused by laser radiation [41]. In Fig. 10.41, a laser ray enters the blown-off plasma from the right and gener­ates a wave in the plasma shown by the curly line. This wave has maxima in plasma density at the vertical bars. The laser ray reflects off these density stripes coher­ently, as if they were a diffraction grating. The reflected ray goes off to the right. The incoming and reflected waves interfere constructively to strengthen the plasma waves, which then reflect more strongly yet. The net result is that much of the incoming light is reflected back toward the laser. Less light reaches the capsule, but that is not the worst part.

Extreme care must be taken to prevent the reflected beam from being amplified as it goes back through the laser. Otherwise, it will fry the laser. There are two kinds of plasma waves that can be generated in a parametric instability. One is an

ion acoustic wave, in which case the instability is called stimulated Brillouin scattering (SBS). The other is an electron plasma wave, in which case the instabil­ity is called stimulated Raman scattering (SRS). The worst part about SRS is that the plasma wave accelerates a beam of electrons. This can preheat the DT fuel so that it cannot be compressed to the required size. All this happens in a very small space, so the beam of electrons is very narrow, forming a pinch. The magnetic field generated by this beam is measured in megagauss (100s of tesla). It is not true that inertial fusion avoids magnetic fields! The instabilities, however, are not those in magnetic fusion.

The higher the frequency of the laser light, the higher are the densities where SBS and SRS occur. Higher frequencies will penetrate more deeply into the plasma corona and minimize these instabilities. This is the reason the NIF laser will use the third harmonic (“3w”) of its fundamental frequency even though almost half the light intensity will be lost in the conversion.