STEM technique maps all the atoms in any molecule

Hank Hogan, hank.hogan@photonics.com and Gary Boas, gary.boas@photonics.com

Transmission electron microscopy (TEM) is a technique that should, in theory, facilitate direct imaging and chemical identification of each and every atom in a material that has an unspecified three-dimensional structure. This is especially true given the recent introduction of aberration-corrected optics. Until recently, however, neither TEM nor any other method has proved up to the task in nonperiodic materials.

Now, however, a group of researchers has devised a technique for imaging and identifying atoms in almost any solid or molecule. The team recently used its creation to resolve and directly identify every atom in a monolayer of boron nitride, a feat never before performed.

The researchers, who included those affiliated with Nion Co. of Kirkland, Wash., Vanderbilt University in Nashville, Oak Ridge National Laboratory and the University of Oxford in the UK, reported in the March 25, 2010, issue of Nature that they had used annular dark-field imaging in an aberration-corrected scanning transmission electron microscope (STEM) in their study.

Using aberration-corrected annular dark-field electron microscopy (left), researchers for the first time have located and identified all atoms in a nonperiodic sample of boron nitride with substitutional impurities. Shown is an experimentally determined model superimposed on the image, with boron (red) and nitrogen (green), along with impurities carbon (yellow) and oxygen (blue). Courtesy of Ondrej L. Krivanek, Nion Co.

This type of atom-by-atom analysis could be performed on organic and other molecules that do not crystallize into an ordered array, said Ondrej L. Krivanek, president of Nion Co. and lead author.

“With the 1-Å – or 100-pm – resolution we now roughly have for light atoms, all the atoms should be identifiable, except probably hydrogen, and the atomic structure of any general molecule should therefore be analyzable in principle,” said Krivanek.

Because electrons are one-fiftieth the size of an atom, they have long been considered prime candidates for direct atomic imaging and identification. However, this has been done only for crystals, which have a regular structure. That has now changed, thanks to various enabling technologies.

One such technology is aberration correction of electron beam imaging, an area that Krivanek said Nion pioneered and in which it is a leader. This has improved imaging resolution by a factor of three. Another advance was the development of very stable electron beams, achieved by eliminating all sources of noise. The result is that the beam used in the current experiments is still to 5 pm rms – 10 times better than is typical.

Other enabling technologies were the aforementioned annular dark-field detection, as well as cold-field emission and ultrahigh-vacuum processing. The first technique, which recently was extended to light atoms, allows chemical identification because the signal is strongly dependent upon atomic number. The second one optimizes resolution when using a lower-energy electron beam, a necessity because these beams minimize radiation damage to structures with light atoms such as boron, carbon, nitrogen and oxygen. Ultrahigh vacuum around the sample prevented stray contaminants from being picked up during the analysis.

Imaging in the study was performed using a Z-contrast UltraSTEM scanning transmission electron microscope made by Nion, with a corrector of third- and fifth-order aberrations and a cold-field emission electron source, which helped to minimize beam blurring.

The aberration corrector and the cold-field electron source gave the researchers excellent resolution at 60 kV. This was important, according to Stephen J. Pennycook, a materials science and technology researcher at Oak Ridge, because operating at this low voltage allowed them to avoid atom displacement damage to the sample.

The use of Z-contrast STEM was integral to the experiment. “The Z-contrast mode is the only way to distinguish the elements based on their intensity in the image,” Pennycook said. Phase contrast imaging is the other common high-resolution TEM mode. Here, however, the atoms are very close in intensity and impossible to distinguish one after the other. Z-contrast imaging uses electrons scattered through larger angles, scattering off the nucleus or Rutherford scattering, and thus is much more sensitive to the species of atom.

“The other big advantage of a Z-contrast microscope,” Pennycook added, “is that it also allows electron energy loss spectroscopy, when transmitted electrons are analyzed for their loss of energy. Although this is a lower-level signal, it is element-specific, so we knew for certain that there was a lot of carbon on the specimen.”

After building and tuning the microscope, the researchers examined boron nitride monolayers. They did this for two reasons, Krivanek explained. One is as a proof of principle of the elemental analysis technique. The second was that boron nitride sheets are potentially useful because the material is related to graphene, which could form the basis for future electronics. Devices could someday be built out of graphene, with nonconducting boron nitride separators.

In their study, the researchers corrected the aberration in an electron beam, swept it across the boron nitride monolayer sample and collected the resulting dark-field image. This revealed individual atoms.

By analyzing the molecular structure of experimental materials, atom by atom, researchers can identify structural defects in those materials. This is significant because defects, including the presence of an impurity atom or molecule, often determine the material’s properties.

They also found atomic substitutions in the monolayer. For example, there was carbon in the place of boron at some locations, carbon in the place of nitrogen at others and oxygen in the place of nitrogen elsewhere. These atoms, which differed in size from the ones they replaced, caused distortions in the monolayer of about one-tenth of an angstrom, or about 10 pm. These observations were in agreement with calculated values.

Although this proves that the technique works, Krivanek noted that there are limitations to keep in mind. The most important is that major radiation damage can occur in a sample, arising from the impact of the electrons in the beam. There are ways around this, particularly if many identical copies of a molecule are present. As that is often the case in biology, the approach could be applicable to the study of organic molecules.

The researchers plan to explore the electron energy loss spectroscopy signal in more detail and hope to correlate the defects observed with electronic changes in the material. This also is very important for graphene, a potential replacement for silicon (which is nearing the famous “end-of-the-road map,” Pennycook said, so next-generation devices will have to use a different material). Graphene offers the right characteristics, but developers have not yet come up with a reliable means to make repeatable devices with it.

“The key is to have the right impurities in the right places and not the wrong ones,” Pennycook said. “So seeing and identifying the impurity atoms is the key to understanding electrical properties and, ultimately, to new devices.”

As for the future, Krivanek said of the technique, “We can now see individual atoms more clearly than we ever could before. Since the entire world is made of atoms, the potential range of applications is very wide indeed.”