The bubbles produced by sonication of liquids are weird things, and there's no doubt that strikingly high temperatures and pressures can be produced in them (for strikingly short periods of time). The idea that they might be high enough to induce nuclear fusion has been around for years, and in 2002 a group at Oak Ridge claimed to have done it. (I wrote about that on my old site here when the news came out, and here a few days later).
In 2004, a further paper from the group came out, followed by another one in January of this year. The lead investigator, Rusi Taleyarkhan, had moved to a new faculty position at Purdue, and a number of people there were working on the idea. I was encouraged. Perhaps I shouldn't have been. Nature is now reporting that many people at Purdue have extreme doubts about Taleyarkhan's work. And if these accounts are accurate, I don't blame them:
Once Taleyarkhan had arrived, lab members became increasingly concerned by his actions. Jevremovic says that he would sometimes examine the equipment and claim that it was producing positive results, referring to an oscilloscope that he had. She says that she was uncertain about how the oscilloscope fitted into the experiment so she asked him for the raw data, but never received any. "He said: 'Look, there's a peak', but there was nothing to see," she says. "I started questioning it."
Then one day, in or soon after May 2004, lab members arrived to find that Taleyarkhan had removed the experimental set-up from the communal lab, and taken it to his own lab off-campus. "I was really upset," says Jevremovic, who had been planning further work with the equipment. . .
Meanwhile, the data in the latest paper have come in for some very hard scrutiny. The experiments are claimed to emit neutrons at the particular energy that would be a sign of nuclear fusion, but it seems that the published data aren't consistent with that at all - and are consistent with those from californium-252, which is a common neutron source. There wasn't supposed to be any such source in these runs, though, which is very disturbing. (A preprint of the paper on these problems is here, for those who want all the details.
I Am Not a Nuclear Physicist, but I am a scientist, and this is all starting to look very sad and suspicious. Depressing as it is to write this, I'm now strongly inclined to think that the Taleyarkhan work is spurious. It would take some amazing results to convince me otherwise, and this would be a mighty good time for him to produce them if he's got 'em. But I don't think he does. As I've said before, the possibility of discoveries like this fills me with joy, which makes it still more painful to watch it all crash in a flaming heap.
The idea of sonochemically induced fusion isn't dead: others are looking at it in different systems. Sonication bubbles are indeed weird (see also here), and much remains to be discovered about them. But a lot of time and effort has seemingly been wasted on the wrong approach, and the whole field ends up looking like a refuge for cranks. It's sad, it's just sad.