Scanning tunneling microscopy sketches exquisite atomic-level landscapes of material surfaces, but reveals little information about the identity of those atoms. Now, two researchers have found a way to pluck small groups of atoms from a surface and tell what they are. The technique should interest the semiconductor industry, which often wants to identify the impurities that can hamper the performance of devices.
The probe of a scanning tunneling microscope has a tip, sharpened to only a few atoms, that hovers several angstroms above the surface to be imaged. Electrons slip or "tunnel" across the gap, and as the probe slides across the surface, the changing current reveals the ridges and valleys of individual atoms. Sometimes a particular atom can be identified from context, but usually it's terra incognita. "There are always these mystery blobs you can't interpret," says John Spence, a physicist at Arizona State University in Tempe.
So Spence and colleague Uwe Weierstall designed a way to identify the anonymous atoms. Their strategy was to apply a 5-volt pulse to the tip to pull the suspect atoms off the surface. They then moved the sample to one side and applied a much stronger pulse--10 kilovolts--which has the opposite effect, flinging the atoms to a detector 20 centimeters away. The flight time from tip to detector reveals the ratio of the atom's charge to its mass, and with a little analysis, the identity of the atoms. (A heavy atom moves more slowly than a light one, as does one with a small charge, which receives a smaller kick from the electric pulse.)
The researchers tested their technique on a silicon surface dotted with islands of germanium. They selectively grabbed the 20-atom islands, and found that most of the travel times indeed corresponded to germanium. But it didn't always work. Some atoms they tested refused to stay on the tip. Spence expects that by cooling the tip with liquid nitrogen, they should be able to reduce thermal noise enough to make the more mobile ones stick.
Mike Miller, a microscopy expert at Oak Ridge National Laboratory in Tennessee is impressed with the technique. "Currently this is the only way to identify atoms on a surface," he says. Miller predicts the Arizona State device could go commercial in 5 to 10 years.