The holy grail of power sources remains a money-hungry and frustratingl y elusive beast. Charles Arthur tracks its progress
Will fusion ever be a viable source of energy? It's a question causing a great deal of turbulence in the fusion community. A proposed new pounds 4 billion fusion reactor, called Iter, would never work, according to recent theoretical research from the US.

The research, by a joint team from the University of Texas and Princeton University, suggests that inside a reactor large enough to generate power by fusion, the random motions of the superheated hydrogen "plasma" would reduce its efficiency so much that Iter (International Thermonuclear Experimental Reactor) "wouldn't work, and by a substantial number", according to Michael Kotschenreuther of the University of Texas.

But that claim has been disputed - "in the past, no theoretical model has effectively predicted the future," said Miklos Porkolab, director of the Plasma Fusion Centre at the Massachusetts Institute of Technology.

While the scientists row, it's clear that the political will to back fusion, which has swallowed up truly tremendous amounts of money in the past 50 years, remains firm. Only last week, the European Commission issued a statement, based on an internal evaluation report, which said that "to maintain Europe's leading position in fusion research, Iter should be built in Europe." It gets better, in political terms - Italy has expressed interest in being the reactor's host country.

The cynic's view of fusion as a power source might borrow a famous economist's remark about the stability of the Brazilian economy: "It lies in the future, and always will do."

Leaving that question aside for a moment, controlled fusion reactors would be the answer to many prayers.

Fusion is the process that powers the sun. The simplest form crashes hydrogen atoms together to produce helium and, crucially, some energy. While the centre of the sun is dense enough and hot enough to power the process using pure hydrogen, a fusion reactor on Earth would have to operate using a mixture of deuterium and tritium - hydrogen atoms with, respectively, one and two neutrons in their nuclei.

At the right temperature and density, the electrostatic repulsion of the hydrogen nuclei is overcome by the "strong nuclear force", and the deuterium nucleus fuses with a tritium nucleus to produce a helium nucleus (two protons and two neutrons), while giving off a neutron and a burst of energy.

In principal, enough energy can be produced to make the reaction self- sustaining: the important thing is to get to a high enough value of what physicists call the "triple product" - the multiple of the temperature, particle density and time - to continue the reaction. The temperature must be between 100 and 200 million degrees Centigrade; the density at least 2 x 1020 particles per cubic metre; and the reaction time at least 1 second. Ignition follows and everyone cracks open the champagne.

So far, nobody has managed that. They are coming closer, though. In 1991, the Joint European Torus (Jet) at Culham, near Oxford, produced a triple product only six times too low to reach ignition. That may sound like a long way off, except that eight years before that, the state of the art rested 700 times away from the magic number. And in reaching that one-sixth figure, the plasma in the Jet generated a 1.7 megawatts for about two seconds - the first time a significant amount of fusion power had been generated in a magnetic confinement device. It was still, however, far short of the amount of energy poured in.

The problem with achieving ignition is that at such high temperatures, the hydrogen turns to plasma and has to be heated and held in an electromagnetic "bottle" by processes which requires vast amounts of energy. Whenever Jet is about to run a major experiment, it has to alert the National Grid, lest lights dim all over Oxfordshire.

The physics of containment, and of the motion of the particles, is so incredibly complex that it has taken 50 years to get to a point where success - a commercial fusion reactor - is still 50 years away. And if the predictions of the turbulence effects are correct, it's even further off. William Dorland, who did some of the new work, told Physics World: "It's good news, bad news and extra good news. The good news is that fusion physicists for the first time really understand something about the process of turbulence. The bad is that the present operating mode for Iter wouldn't meet expectations by a large margin. The extra good news is that once physicists understand a phenomenon, they can exploit that to make the machine better."

If fusion could work, it has huge advantages over fission, and indeed most other electricity fuels. The sources won't run out soon. The lifetime electricity requirement of the average person in an industrialised country could be provided by 10 grams of deuterium (which can be extracted from 500 litres of water) and 15g of tritium (produced from 30g of lithium, which is plentiful in the Earth's crust).

It wouldn't contribute to the greenhouse effect or other atmospheric pollution. Malfunction would lead to a shutdown, rather than a "runaway". Finally, while spare neutrons will create some radioactivity in the reactor structure, it will have a short half-life, and won't require the geological timespans of disposal required for fission waste.

But fusion, or at least ignition, remains an incredibly elusive target. Europe remains in the lead in research, having in the past 10 years put eight billion ecu (about pounds 10.6 billion) into projects such as Jet. "Jet will continue until the end of this century and maybe into the next," says Martin Keilhacker, director of Jet, where another series of experiments will begin in a few weeks' time, after rebuilding last year. He says that approaching ignition is a process of diminishing returns; to achieve it would require a reactor twice as big in each dimension. That means it would have a volume eight times bigger than Jet's; the cost would probably be 10 times greater.

Iter, if it works, would aim eventually to burn hydrogen for about an hour; a subsequent reactor would aim for continuous operation. The step after that would be a commercial fusion reactor - but that, says Professor Keilhacker, is "probably 50 years or so from now".

This raises the question - how can we be sure that the political will to build fusion reactors will remain? "It is politically difficult," says Professor Keilhacker. "The timescale goes beyond that of physicists or engineers - and especially of politicians."