Fusion power

The terms "fusion experiment" and "fusion device" refer to the collection of technologies used for scientific investigation of plasma, and technical advancement. Not all are capable of, or routinely used for, producing thermonuclear reactions i.e. fusion.

The term "fusion reactor" is used interchangeably to mean the above experiments, or to mean a hypothetical power-producing version, at the center of a commercial power plant, requiring additions such as a breeding blanket and heat engine.

Fusion reactions occur when two or more atomic nuclei come close enough for long enough that the nuclear force pulling them together exceeds the electrostatic force pushing them apart, fusing them into heavier nuclei. For nuclei heavier than iron-56, the reaction is endothermic, requiring an input of energy. The heavy nuclei bigger than iron have many more protons resulting in a greater repulsive force. For nuclei lighter than iron-56, the reaction is exothermic, releasing energy when they fuse. Since hydrogen has a single proton in its nucleus, it requires the least effort to attain fusion, and yields the most net energy output. Also, since it has one electron, hydrogen is the easiest fuel to fully ionize.

The repulsive electrostatic interaction between nuclei operates across larger distances than the strong force, which has a range of roughly one femtometer—the diameter of a proton or neutron. The fuel atoms must be supplied enough kinetic energy to approach one another closely enough for the strong force to overcome the electrostatic repulsion in order to initiate fusion. The "Coulomb barrier" is the quantity of kinetic energy required to move the fuel atoms near enough. Atoms can be heated to extremely high temperatures or accelerated in a particle accelerator to produce this energy.

An atom loses its electrons once it is heated past its ionization energy. The resultant bare nucleus is a type of ion. The result of this ionization is plasma, which is a heated cloud of bare nuclei and free electrons that were formerly bound to them. Plasmas are electrically conducting and magnetically controlled because the charges are separated. This is used by several fusion devices to confine the hot particles.

A reaction's cross section, denoted σ, measures the probability that a fusion reaction will happen. This depends on the relative velocity of the two nuclei. Higher relative velocities generally increase the probability, but the probability begins to decrease again at very high energies.

In a plasma, particle velocity can be characterized using a probability distribution. If the plasma is thermalized, the distribution looks like a Gaussian curve, or Maxwell–Boltzmann distribution. In this case, it is useful to use the average particle cross section over the velocity distribution. …

Plasma is an ionized gas that conducts electricity. In bulk, it is modeled using magnetohydrodynamics, which is a combination of the Navier–Stokes equations governing fluids and Maxwell's equations governing how magnetic and electric fields behave. Fusion exploits several plasma properties, including:

Many approaches, equipment, and mechanisms are employed across multiple projects to address fusion heating, measurement, and power production.

A deep reinforcement learning system has been used to control a tokamak-based reactor. The system was able to manipulate the magnetic coils to manage the plasma. The system was able to continuously adjust to maintain appropriate behavior (more complex than step-based systems). In 2014, Google began working with California-based fusion company TAE Technologies to control the Joint European Torus (JET) to predict plasma behavior. DeepMind has also developed a control scheme with TCV. Researchers at Princeton Plasma Physics Laboratory (PPPL) and Princeton University made significant advances using artificial intelligence to control plasma in a tokamak, achieving high confinement without harmful edge energy bursts (known edge-localized modes) on two tokamaks: DIII-D and KSTAR. An international collaboration involving some of the same researchers from Princeton University and PPPL, as well as researchers from Chung-Ang University, Columbia University and Seoul National University, also led to a new AI system known as Diag2Diag, that fills in missing sensor data for fusion systems, offering more detail than a real-world sensor could have provided. Data gathered from the Diag2Diag research also provided new support for a leading theory about controlling plasma disruptions. AI agents for designing inertial fusion energy systems have been developed by a joint team from Lawrence Livermore National Laboratory, Los Alamos National Laboratory, and Sandia National Laboratories.

The diagnostics of a fusion scientific reactor are extremely complex and varied. The diagnostics required for a fusion power reactor will be various but less complicated than those of a scientific reactor as by the time of commercialization, many real-time feedback and control diagnostics will have been perfected. However, the operating environment of a commercial fusion reactor will be harsher for diagnostic systems than in a scientific reactor because continuous operations may involve higher plasma temperatures and higher levels of neutron irradiation. …

The fuels considered for fusion power are mainly the heavier isotopes of hydrogen—deuterium and tritium. Deuterium is abundant on earth in the form of semiheavy water. Tritium, decaying with a half-life of 12 years, must be produced. Fusion reactor concepts assume as a component a proposed lithium "breeding blanket" technology surrounding the reactor. Helium-3 is a more speculative fuel, which must be mined extraterrestrially or produced by other nuclear reactions. The protium–boron-11 reaction is extremely speculative, but minimizes neutron radiation.

The easiest nuclear reaction, at the lowest energy, is D+T:

This reaction is common in research, industrial and military applications, usually as a neutron source. Deuterium is a naturally occurring isotope of hydrogen and is commonly available. The large mass ratio of the hydrogen isotopes makes their separation easy compared to the uranium enrichment process. Tritium is a natural isotope of hydrogen, but because it has a short half-life of 12.32 years, it is hard to find, store, produce, and is expensive. Consequently, the deuterium-tritium fuel cycle requires the breeding of tritium from lithium using one of the following reactions:

The reactant neutron is supplied by the D–T fusion reaction shown above, and the one that has the greatest energy yield. The reaction with Li is exothermic, providing a small energy gain for the reactor. The reaction with Li is endothermic, but does not consume the neutron. Neutron multiplication reactions are required to replace the neutrons lost to absorption by other elements. Leading candidate neutron multiplication materials are beryllium and lead, but the Li reaction helps to keep the neutron population high. Natural lithium is mainly Li, which has a low tritium production cross section compared to Li so most reactor designs use breeding blankets with enriched Li.

Drawbacks commonly attributed to D–T fusion power include:

The neutron flux expected in a commercial D–T fusion reactor is about 100 times that of fission power reactors, posing problems for material design. …