The Idea of Sonofusion (technically known as acoustic inertial confinement fusion) was derived from a related phenomenon, sonoluminescence, which has been known for more than 70 years. In sonoluminescence, a process widely used by chemists and also a science-fair staple, loudspeakers attached to a liquid-filled flask send pressure waves through the fluid, exciting the motion of tiny gas bubbles. The bubbles periodically grow and collapse, producing visible flashes of light that last less than 50 picoseconds.

About 20 years ago, researchers studying these light-emitting bubbles speculated that their interiors might reach such high temperatures and pressures that they could trigger fusion reactions. Since then, several groups have been trying to achieve fusion using sound waves, most of them with a kind of enhanced sonoluminescence. This method, called single-bubble sonoluminescence, involves a single gas bubble that is trapped inside the flask by a pressure field and yields light flashes during repetitive implosions.

Our own efforts began in 1996 at the Oak Ridge National Laboratory in Tennessee [see photo, "Bubble Maker"]. In those first attempts, we tried many different configurations for single-bubble sonoluminescence, all without success. We finally concluded that excitation pressures higher than about 170 kilopascals would always dislodge the bubble from its stable position and disperse it in the liquid. That, we thought, was a fundamental problem for single-bubble sonoluminescence, because we calculated we would need at least 10 times that pressure level to implode the bubbles strongly enough to trigger thermonuclear fusion.

To overcome that limitation, we began seeking a different approach. After a lot of brainstorming and many experiments, we came up with a promising idea: remove virtually all the naturally occurring gas bubbles dissolved in the liquid and then, somehow, create our own bubbles, much smaller, precisely when we needed them. That way, we could increase the bubbles' maximum size before their collapse, thereby tremendously increasing the implosion's energy. We easily removed the gas from the liquid by attaching a vacuum pump to the flask and acoustically agitating the liquid. But then how could we create the bubbles we needed? We owe our success—as often occurs in science—to some fortunate happenstance, perhaps more than we had any right to expect.

The idea came from one of our colleagues, who was then working on a largely unrelated project. The colleague was trying to use a sonoluminescence flask as a neutron detector. To test his idea, he fired high-energy neutrons at the flask and then analyzed the light emissions. Upon learning about that, we figured we could do the same thing, not to produce light but to create tiny vapor bubbles that we could later grow and implode.

Our apparatus has evolved since those first experiments in 1996, but it continues to be relatively simple [see illustration, ""HowSonofusionWorks.pdf""]. It consists of a cylindrical Pyrex glass flask 100 millimeters high and 65 mm in diameter. We first attach a lead-zirconate-titanate ceramic piezoelectric crystal in the form of a ring to the flask's outer surface. This piezoelectric ring works like the loudspeakers in a sonoluminescence experiment, although it creates much stronger pressure waves. When a positive voltage is applied to the piezoelectric ring, it contracts; when the voltage is removed, it expands to its original size.

We then fill the flask with commercially available deuterated acetone, in which 99.9 percent of the hydrogen atoms in the acetone molecules are deuterium (this isotope of hydrogen has one proton and one neutron in its nucleus). The main reason we chose deuterated acetone is that atoms of deuterium can undergo fusion much more easily than ordinary hydrogen atoms. Also, the deuterated fluid can withstand significant tension ("stretching") without forming unwanted bubbles. The substance is also relatively cheap, easy to work with, and not particularly hazardous.

To initiate the sonofusion process, we apply an oscillating voltage with a frequency of about 20 000 hertz to the piezoelectric ring. The alternating contractions and expansions of the ring—and thereby of the flask—send concentric pressure waves through the liquid. The waves interact, and after a while they set up an acoustic standing wave that resonates and concentrates a huge amount of sound energy. This wave causes the region at the flask's center to oscillate between a maximum (1500 kPa) and a minimum (-1500 kPa) pressure. During the positive pressure cycle, the liquid is being compressed, and during the negative pressure cycle, it is being stretched.

Precisely when the pressure reaches its lowest point, we fire a pulsed neutron generator, a commercially available, baseball bat-size device that sits next to the flask. The generator emits high-energy neutrons at 14.1 mega-electronvolts in a burst that lasts about 6 microseconds and that goes in all directions. Some neutrons go through the liquid, and some collide head-on with the carbon, oxygen, and deuterium atoms of the deuterated acetone molecules. In these collisions, the fast-moving neutrons may knock the atom's nuclei out of their molecules. As these nuclei recoil, they give up their kinetic energy to the liquid molecules. This interaction between the nuclei and the molecules creates heat in regions a few nanometers in size that results in tiny bubbles of deuterated acetone vapor. Our experiments, along with computer simulations, suggest that this process generates clusters of about 1000 bubbles, each with a radius of only tens of nanometers.

By firing the neutron generator during the liquid's low-pressure phase, the bubbles instantly swell—a process known as cavitation. In this swelling phase, the bubbles balloon out 100 000 times from their nanometer dimensions to about 1 mm in size. To grasp the magnitude of this growth, imagine that the initial bubbles are the size of peas. After growing by a factor of 100 000, each bubble would be big enough to contain the Empire State Building. Then, as the pressure cycle rapidly reverses, the liquid pushes the bubbles' walls inward with tremendous force, and they implode with great violence.

The implosion creates spherical shock waves within the bubbles that travel inward at high speeds and significantly strengthen as they converge to their centers. The result, in terms of energy, is extraordinary: our hydrodynamic shock-wave computer simulations show that the shock waves create, in a small region at the center of the collapsing bubble, a peak pressure greater than 10 trillion kPa. For comparison, the atmospheric pressure at sea level is 101.3 kPa. The peak temperature in this tiny region soars above 100 million degrees centigrade, about 20 000 times that of the sun's surface.

These extreme conditions within the bubbles—especially the bubbles at the center of the cluster, where the shock waves are more intense because of the surrounding implosions—cause the deuterium nuclei to collide at high speed. These collisions are so violent that the positively charged nuclei overcome their natural electrostatic repulsion and fuse. The fusion process creates neutrons, which we detected using a scintillator, a device in which radiation interacts with a liquid that gives off light pulses that can be measured. The process is also accompanied by bursts of photons, which we detected with a photomultiplier. And subsequently, after about 20 microseconds, a shock wave in the liquid reaches the flask's inner wall, resulting in an audible "pop," which can be picked up and amplified by a microphone and a speaker.

Increasing the pressure by an order of magnitude, firing neutrons at the flask to seed the bubbles on demand, and choosing a liquid rich in deuterium are the three key differences between single-bubble sonoluminescence and our sonofusion method.