Scientists have long dreamed of tapping the power of nuclear fusion, the process that keeps stars burning. Today's best attempts, using magnetic fields in hugely expensive experimental reactors, are promising but so far have not yielded sizable amounts of energy on a sustained basis. So, in 2002, when researchers reported a new way of achieving fusion by imploding bubbles in a liquid, hopes ran high that the technique could somehow be scaled up eventually to provide an alternative way of generating electricity.

The researchers, led by Rusi P. Taleyarkhan, then at Oak Ridge National Laboratory, in Tennessee [see photo, " "], described in a 2002 paper in Science how they had blasted a glass flask filled with a liquid rich in deuterium, a hydrogen isotope, with high-frequency sound, creating pressure oscillations that imploded tiny bubbles in the liquid. The bubbles' violent collapse, the researchers believed, caused some of the deuterium to undergo fusion. Their novel method, dubbed bubble fusion or sonofusion, was controversial from the beginning, and so far no independent group has been able to replicate the experiment.

Many experts say they do believe that making hydrogen nuclei fuse in imploding bubbles is possible in principle--and in this respect the bubble fusion controversy is fundamentally different from the "cold fusion" debate that embroiled the physics community in 1989. But those experts hasten to add that they are not convinced Taleyarkhan actually achieved it. Their main concern is whether Taleyarkhan's group has ruled out possible sources of errors in the tricky business of detecting the neutrons that are a characteristic fusion signature.

Taleyarkhan and his collaborators addressed this and other criticism and reported continued progress in the March 2004 issue of Physical Review E and in the January 2006 issue of Physical Review Letters , which are among the top journals in their field. And in May a year ago, this magazine published a description of their work for the general reader.

But now controversy is bubbling again. This past March, Purdue University, in West Lafayette, Ind., announced that it was initiating a formal review of the bubble fusion research by Taleyarkhan, who is now a full professor at the university. Purdue launched the review, which will report its initial findings by 1 June, after concerns about Taleyarkhan's work were brought to the university's provost by other nuclear engineering faculty members who were trying to replicate his experiment and complained that he was uncooperative and secretive.

Taleyarkhan told IEEE Spectrum that he was surprised by the allegations, which he said had not been discussed with him directly, and that he stands by his work.

Concerns about Taleyar-khan's work at Purdue broke into the open in early March with a report in Nature magazine. According to that report, Lefteri Tsoukalas, head of the nuclear engineering school at Purdue, along with other faculty members, claimed that Taleyarkhan had refused to provide raw experimental data, removed equipment from a shared laboratory, and opposed publication of negative results obtained in some of their experiments.

To that, Taleyarkhan says that he was always willing to help and that the results obtained were actually positive, as reported by two Purdue graduate students working with Tsoukalas in the May 2005 issue of Nuclear Engineering and Design . Taleyarkhan's critics at Purdue, however, insist that he influenced that work and therefore the results don't serve as confirmation of bubble fusion.

Meanwhile, a group of researchers specifically funded by the U.S. Defense Advanced Research Projects Agency (DARPA) to reproduce Taleyarkhan's results saw no evidence of fusion either. This effort, in which Taleyarkhan also participated, took place mostly at the laboratory of Seth Putterman, who also has been seeking ways to achieve fusion in collapsing bubbles, at the University of California, Los Angeles.

Though Taleyarkhan and his collaborators are able to provide lucid accounts of how they believe they've achieved bubble fusion, relying on accepted principles of nuclear physics, skepticism centers on whether their neutrons are truly fusion's telltale neutrons. One problem immediately identified by critics was that Taleyarkhan was using high-energy neutrons from an external generator to initiate bubble formation in the liquid: were the neutrons allegedly produced by fusion actually left over from the neutron generator?

In his latest setup, Taleyarkhan dealt with that issue by eliminating the neutron generator; he now "seeds" the bubbles instead with natural uranium dissolved in the liquid. But even now, experts aren't satisfied that the observed neutrons are really from fusion. They could actually come from spurious sources, such as cosmic rays or radio-active contamination, says Fred Becchetti, a physics professor at the University of Michigan, in Ann Arbor.

In a commentary submitted to Physical Review Letters , Brian Naranjo, a graduate student in Putterman's laboratory, analyzed data published in Taleyarkhan's latest paper and concluded that the energy spectrum presented as coming from neutrons produced in fusion is not the one expected for that type of reaction. Naranjo says that instead the spectrum is a much better match for neutrons produced by the radioactive decay of californium-252, a material commonly used in nuclear engineering laboratories.