A fusion reaction is one in which two atomic nuclei merge to form a heavier nucleus and, in most cases, an accompanying product such as a free nucleon. In almost all fusion reactions between light nuclei, a portion of their rest mass is converted into kinetic energy of the reaction products, or into gamma rays. Stars produce energy through various fusion reactions. In stars such as the Sun, the net effect is to convert hydrogen nuclei ( protons) into helium nuclei. The released kinetic energy and gamma rays heat the stellar interior, maintaining it at the high temperatures (greater than 10 million K) required to continue the fusion. The thermal energy of the nuclei drives them together despite their electrostatic repulsion. Such conditions are called thermonuclear.

Scientists have long worked toward the goal of using thermonuclear fusion reactions to produce useful power. The most promising two reactions involve the isotopes of hydrogen: deuterium (one proton and one neutron) and tritium (one proton and two neutrons). Deuterium occurs naturally as a minor constituent in hydrogen-containing materials, in quantities sufficient to meet the energy needs of societies for billions of years. Tritium can be bred from lithium by a neutron-induced reaction in a blanket that could conceivably surround a fusion reactor. The western United States contains large lithium deposits in the salts of dry lake beds, and much larger quantities are dissolved in the ocean.

The reaction that occurs with the greatest probability and at the lowest temperatures involves fusing a deuterium nucleus with a tritium nucleus to form a helium (He4) nucleus, along with an energetic neutron that helps sustain further fusion. The products contain 17.6 million electron volts (MeV) of released kinetic energy. (The hydrogen bomb employs this as a brief, uncontrolled reaction.) The second reaction, the fusing of two deuterium nuclei, has two branches that occur with about equal probability. One leads to an He3 nucleus, a neutron, and 3.2 MeV of kinetic energy. The other produces a tritium nucleus, a proton, and 4.0 MeV. While this could furnish power for nearly the Sun's lifetime, it is the somewhat more easily produced deuterium-tritium reaction — itself sufficient for many thousands of years — that will provide most of the energy in the next generation of research devices.

Fusion reactions are easily induced by using a charged particle accelerator) to bombard a solid or gaseous tritium target with energetic deuterium nuclei. However, this technique consumes rather than produces power, because most of the accelerated nuclei lose their energy through elastic collisions with electrons and nuclei, without producing fusion reactions. A net energy gain is obtained only by mimicking the thermonuclear conditions of stars, but because a reactor is small and must operate in a limited time frame, it must actually have a much higher power density and be several times hotter than a stellar core. The advantage of carrying out the reactions under thermonuclear conditions is that the energy lost by one nucleus in an elastic collision is transferred to the particle it hits and is still available to initiate a fusion reaction.

At thermonuclear temperatures, matter can exist only in the primarily ionic plasma state. Fusion reactions occurring in a plasma heat it further, because the portion of the reaction energy that remains with the electrically charged reaction products is transferred to the bulk of the plasma through collisions. In the deuterium-tritium reaction, the positively charged helium nucleus carries 3.5 MeV. The neutron escapes the plasma with little interaction and, in a reactor, could deposit its 14.1 MeV in a surrounding lithium blanket. This would breed tritium and also heat an exchange medium (such as helium), which could be used to produce steam to turn generator turbines. However, the plasma loses thermal energy through conduction, convection, and bremsstrahlung (the electromagnetic radiation emitted when a charged particle decelerates). Energy also escapes through electrons undergoing level transitions in heavier impurities, and through losses of hot nuclei that capture an electron and escape any confining fields. Ignition occurs only when the energy deposited equals or exceeds the energy being lost.

A plasma at millions of degrees is not compatible with an ordinary confining wall, but the effect is not the destruction of the wall, as might be expected. Although the temperature of a thermonuclear plasma is very high and the power flowing through it may be large, the stored energy is relatively small and would quickly be radiated away by impurities if the plasma touched a wall and began to vaporize it. Any significant contact with the vessel housing it causes the plasma's extinction within a few thousandths of a second.

Magnetic Confinement
Since the early 1950s, most fusion research has used magnetic fields to confine the charged particles constituting a plasma. The density required for this is much lower than atmospheric density, so the plasma vessel is evacuated, then filled with the hydrogen-isotope fuel at 0.000001 times that density. Magnetic-field configurations fall into two types: open and closed. In an open configuration the charged particles, which are spiraling along magnetic field lines maintained by a solenoid, are reflected at each end of a cell by stronger magnetic fields. In this simplest type of "mirror machine," many particles that have most of their velocity parallel to the solenoidal magnetic field are not reflected and can escape. Present-day mirror machines retard this loss by using additional plasma cells to set up electrostatic potentials that help confine the hot ions within the central solenoidal field.

In closed configurations, the magnetic-field lines along which charged particles move are continuous within the plasma. This closure has most commonly taken the form of a torus, or doughnut shape. The most common example is the tokamak. In this device the primary confining field is toroidal and is produced by coils surrounding the vacuum vessel. Other coils cause current to flow through the plasma by induction. This toroidally flowing current engenders a poloidal magnetic field, at right angles, that wraps itself around the plasma. The poloidal field and the stronger toroidal field, acting together, yield magnetic-field lines that spiral around the torus. This spiraling ensures that a particle spends equal amounts of time above and below the toroidal midplane. This cancels the effects of a vertical drift that occurs because the magnetic field is stronger on the inside of the torus than on the outside.

Plasma Heating
Tokamak plasmas can be heated to temperatures of 10–15 million K by the current flowing in the plasma. At higher temperatures the plasma resistance becomes too low for this method to be effective, and heating is accomplished by injecting beams of very energetic neutral particles into the plasma. These ionize, become trapped, and transfer their energy to the bulk plasma through collisions. Alternatively, radio-frequency waves are launched into the plasma at frequencies that resonate with various periodic particle motions. The waves give energy to these resonant particles, which transfer it to the rest of the plasma through collisions.

Current Drive
Experiments are also under way in which radio-frequency waves are used to push electrons around the tokamak to maintain the plasma current. Such noninductive current drive allows the tokamak pulse to outlast the time limits imposed by the fact that, in a transformer-driven tokamak, the plasma current lasts only as long as the current in the secondary coils is changing. When the secondary coils reach their current limits, confinement is lost, and the plasma terminates until the transformer can be reset (a matter of at least seconds). Although the plasma in an inductively driven tokamak is pulsed, the electricity produced would not be, because the thermal inertia of the neutron-capturing blanket would sustain steam generation between pulses. By allowing longer pulse or steady-state plasma operation, however, radio-frequency current drive could lessen the thermal stresses.

Inertial Confinement
Another approach to fusion, termed inertial confinement, aims to compress a capsule of deuterium and tritium to very high temperatures and densities in a process analogous to what occurs in a hydrogen bomb. The compression is accomplished by bombarding the capsule from all sides at once with an intense pulse of laser light, ions, or electrons. The outer pellet mass vaporizes and imparts inwardly directed momentum to the remaining pellet core. The inertia of the inwardly driven material must be sufficient to localize the resulting fusion plasma for the approximately 10 -9 sec required to get significant energy release. In recent years, the inertial confinement research program has concentrated on the use of X rays to compress the fuel capsule. The X rays are produced by very high power lasers or ion beams striking a high-atomic-number cylinder surrounding the capsule. The advantage of this indirect drive is that much less spatial symmetry is required of the primary driver beams.

Progress toward Energy Production
The minimum confinement condition needed to achieve energy gain in a deuterium-tritium plasma — the so-called Lawson criterion — is that the product of density in ions per cm3 and energy containment time in seconds must exceed 6 X 1013. This was first attained in 1983 in a hydrogen plasma at the Massachusetts Institute of Technology (MIT). In 1986 the Tokamak Fusion Test Reactor (TFTR) at Princeton University's Plasma Physics Laboratory demonstrated the existence of the theoretically predicted "bootstrap current," which arises from a dynamo effect within the plasma. This means that most of the electric current needed to sustain the poloidal component of the magnetic field perhaps can be supplied at little or no cost by the plasma itself. Beginning in 1993, the TFTR became the first fusion device to run routinely with a deuterium-tritium fuel mixture, for which the energy confinement proved significantly better than for a deuterium-only plasma. In 1994 the amount of power released in its pulses reached 10.7 million watts, and the temperatures reached were more than 40 times those at the Sun's core.

If simple scattering by collisions between isolated pairs of particles were the only mechanism by which energy was lost, then achieving the quality of confinement necessary for an electricity-producing fusion reactor would be a daunting but straightforward undertaking. The reality has been that early fusion experiments encountered rates of energy leakage too large to permit success. Progress occurred during the 1970s and 1980s as successive generations of tokamaks led to a better understanding of how the excess leakage, called "anomalous" diffusion, could be reduced by varying such factors as the dimensions of the plasma torus and the magnitude of the electric current flowing within it, and by finding better modes of operation. In the 1990s physicists began intensive explorations of the possibility of further improving confinement by carefully tailoring the distribution across the plasma of the electric current flowing within it. It was discovered in 1995 that by reversing the change in pitch, or shear, of the helical magnetic-field lines that hold the plasma, a core region could be created inside the shear reversal in which the unwanted leakage of energy and particles was reduced to very low levels.

Much research has yet to be done, however, before fusion power reactors become a reality. Japan, Russia, the European Union (EU), and the United States agreed in 1992 to work on the design of an International Tokamak Experimental Reactor capable of reaching ignition in the early part of the 21st century. In the meantime, South Korea is building a tokamak with superconducting coils that will be capable of maintaining high-temperature plasmas for pulses of up to 300 seconds. With the completing of the TFTR experiments at Princeton University in 1997, work is proceeding there on a spherical tokamak, to study the reactor-relevant physics of low-aspect-ratio plasmas (toroids for which the ratio of the major to minor radii is significantly less than 2). Work also continues on tokamaks at General Atomics in San Diego, Calif., and at MIT in Cambridge, Mass. Japan and the EU each have a large tokamak as well as smaller confinement devices.

The goal of fusion is, in effect, to make and hold a small star. Such a goal is daunting. That it is pursued even so is an indication of the benefits success could bring. Besides providing an almost inexhaustible fuel supply, fusion is environmentally benign. The resulting ash is harmless helium, the afterheat in the reactor structure would be much less than in a fission reactor, and the heat would be distributed through a greater thermal mass. Also, because fusion is not a chain reaction, it cannot run out of control. Any perturbation would cause the plasma to extinguish itself. It would also be difficult to produce nuclear-weapons materials surreptitiously at a fusion plant. Because no fissionable material should ordinarily be present, it would be a simple matter to detect characteristic gamma rays. Present levels of support for research are aimed at building a demonstration fusion power plant early in the 21st century.

Larry R. Grisham

Bibliography: Conn, R., et al., "The International Thermonuclear Experimental Reactor," Scientific American, April 1992; Fowler, T. K., The Fusion Quest (1997); Fujiwara, M., and Wan, Y., eds., The Frontier of Physics in Fusion-Relevant Plasmas (1998); Glanz, J., "Common Ground for Fusion," Science, Aug. 6, 1999; Panarella, E., ed., Current Trends in International Fusion Research (1997); Riordon, J., "Fusion Power from a Floating Magnet?" Science, Aug. 6, 1999; Roth, J., Introduction to Fusion Energy (1986); Szirmay, L. V., Nuclear Fusion Reactor Design Fundamentals (1992); Velard, G., et al., eds., Nuclear Fusion by Inertial Confinement (1993); Wesson, J., and Campbell, D., Tokamaks, 2d ed. (1997).