In the current version, a round gallium arsenide sensor chip outputs a pulse for every 13th radioactive particle it encounters. A model on the drawing board would sense every fifth thermal neutron and cost roughly $10.

"My ultimate goal is to increase the efficiency of our detector to 30 percent. At that level, we will be seriously competing with all the different types of detectors that are out there already, except that ours will only cost between $10 and $40," said professor Douglas McGregor at Kansas State University here. McGregor co-invented the diode with Raymond Klann, a researcher at the U.S. Department of Energy's Argonne National Laboratory (Argonne, Ill.).

Radioactive materials have excess neutrons in their nuclei, making their isotope state heavier than their normal state. Over time the extra neutrons are expelled by the radioactive isotope as high-energy particles - thermal neutrons. The high-energy neutrons irreversibly damage the body's tissues, and the damage from exposure is cumulative. To manage the problem, personnel working in proximity to nuclear materials wear dosimeters - badges that must be sent in for processing to determine the level and duration of exposure.

Such a non-real-time approach was acceptable when it was assumed that any accidental exposure would be low-level. But the threat of nuclear terrorism has raised the possibility of quick, higher-level overexposures from such sources as "dirty" bombs, contamination of the mail or even contamination of water and air supplies. The only way to monitor levels in real-time today is to use a Geiger counter the size of a lunch box, but such instruments are typically designed to be used under the supervision of trained personnel and do not monitor personal exposure in the way that a dosimeter does. Geiger counters depend on a column of gas that is about a foot long and an inch in diameter, and they require a high-voltage power source of up to 5,000 volts, making them bulky and expensive.

"We want to make it easier to detect weapons of mass destruction - in particular, nuclear weapons. When fully developed, our neutron detectors could assist international weapons inspections in Iraq or prevent entry of nuclear materials at our borders," said McGregor.

The McGregor-Klann GaAs-based neutron detectors have been nine years in the making. The long lead time wasn't considered a problem until late last year. "We are getting a lot more interest from industry and the national labs since Sept. 11," McGregor said. While Klann at Argonne National Labs has secured most of the project's funding to date, McGregor said, "Now other National Laboratories are interested, and even some commercial companies are starting to feel the urgency to develop cheap, real-time nuclear detectors."

Efficiency strides

The upside of traditional gas-column detectors is that they are almost 100 percent efficient, detecting virtually every single emitted neutron. Solid-state detectors, by contrast, have always been low-efficiency and hence have never been commercialized. The typical 0.5 percent efficiency of solid-state detectors would require 200 neutrons for every pulse, making them too insensitive for practical use.

But McGregor and Klann have slowly improved the efficiency of their design from 0.5 percent to more than 13.5 percent for the current prototype. The next goal is 20 percent efficiency - not quite the 30 percent that has been the grail for solid-state sensors, but it's probably good enough to address the new terrorism threats.

"There are many terrorist threats that our detectors could alleviate," McGregor said. "For instance, we don't want our stockpiles of nuclear materials and weapons falling into terrorists' hands, so if we put detectors within [those items'] vicinity we could monitor any change in the amount of neutrons being emitted in real-time. Any alteration in the expected number could set off an alarm and prevent the theft."

Over the years, the two scientists have developed a portfolio of detectors, ranging from models requiring up to 50 V of external power down to a fully self-powered model. What all have in common is a GaAs architecture that converts impinging thermal neutrons into pulses that can be counted.

"The power supply requirements range from 50 V down to 1 V, which we use to reverse-bias the diode. But I have built some detectors that are actually self-biased," said McGregor. "They operate from their own internal potentials, so you need not apply any external voltage to them at all."

The diode is virtually the entire device, covering most of the 6-mm diameter of the detector. The design starts with a grounded n-type GaAs substrate with a high-purity active GaAs layer. Next is a p+ Schottky barrier layer, which is connected to an integrating preamplifier. This structure is the surface diode that conducts the pulses generated by impinging neutrons.

The layer that turns the diode into a detector is a neutron-reactive boron coating (some models use lithium instead of boron) on the surface diode. When a thermal neutron hits the boron film, it is absorbed and splits a boron nucleus into charged particles. One of the particles, an alpha particle (a high-speed helium nucleus), causes a pulse to be produced by the GaAs diode when it excites millions of electrons. The electrons form a diode current, which is integrated into a pulse.

As is the case with Geiger counters, there is one pulse per neutron. Pulses can be counted and displayed on a lapel-sized readout after being "normalized" by dividing by the efficiency. "We can make anywhere between 50 and 100 of these devices on a wafer," said McGregor.

Millions of holes

The inventors improved the efficiency of the detector via micromachining techniques, rather than a new architecture or new materials formulations. "The problem we encountered was that film stress limited the amount of boron film we could put on the diode, thereby limiting the efficiency of the device," McGregor said. "We finally solved that problem by etching millions of tiny holes - on the order of 3 microns in diameter - over the surface of the diode. Now, when we apply a thick, multicoated film, we get a 27x increase in efficiency."

McGregor said he knows of one lab that has published a design with 7 percent efficiency, but he is confident that his team's nine years of experimentation will keep it ahead of the competition. The team's next-generation device will vary the configuration of the perforations on the top layer of the diode in an attempt to increase the efficiency further.

"We have some ideas about how to get to 30 percent in just a few years," said McGregor. By then, he predicts, some early adopters will be designing nuclear detectors using the team's GaAs sensor chips.

Eventually, he predicts a very thin, compact badge, small enough to be worn on a lapel, would marshal a little circuitry and a watch battery to read out the exact number of thermal neutrons encountered by the wearer.