Microscopy Method Goes Deep

As a teenager, David N. ­Seidman was so fascinated with atoms that he longed to be able to look at them. “But my high school chemistry teacher said it was impossible to see atoms,” recalls Seidman, now a professor of materials science and engineering at Northwestern University.

Little did Seidman’s teacher know, atoms were about to show themselves for the first time. Fifty-nine years ago this week, Pennsylvania State University physics professor Erwin W. Müller made history by directly imaging individual ­atoms with an inexpensive instrument he designed and built: a field-ion microscope. Inside the evacuated glass apparatus, a powerful electric field ran through a sharply pointed tungsten specimen, causing atoms to fly off the surface of the metal tip and reveal themselves by lighting up a nearby fluorescent screen.

Seidman did eventually get to “see” atoms by using field-ionization methods as a young faculty member at Cornell University in the 1960s. Today, he and other atom-loving researchers carry on Müller’s work by using atom-probe tomography (APT), a direct descendant of field-ion microscopy, to examine the building blocks of matter.

Like the older method, APT provides single-atom sensitivity. But the modern technique is more powerful: Not only can it determine, with subnanometer resolution, the three-dimensional atomic structure of internal or buried interfaces, it can also determine the chemical identity of the atoms in samples. Understanding the atomic structure deep within a material is important because it dictates that material’s properties. The position of atoms at the interface of two materials inside a semiconductor device, for instance, often controls its function.

Despite APT’s analytical prowess and more than a dozen years of instrument commercialization, the technique’s reach and popularity have grown slowly. In just the past few years, however, the number of instruments, practitioners, scholarly journal papers, and classes of materials analyzed by APT has increased by leaps and bounds.

Once limited mainly to probing metals and alloys, APT now exposes hidden 3-D nanostructures in oxides, semiconductors, biological specimens, and other classes of materials that were previously inaccessible to APT and microscopy methods. With its expanded reach, the method is giving scientists a deeper understanding of materials phenomena critical to a host of topics ranging from technology applications, such as catalysts, batteries, and microelectronics, to basic biology, geology, and planetary science.

Innovations in sample preparation and APT instrumentation get much of the credit for the technique’s recent expansion to new materials. With regard to sample preparation, field-ionization aficionados such as Seidman, one of the field’s longest-practicing researchers, note that a key challenge in doing this type of analysis is making the requisite needlelike specimens.

As was the case in Müller’s day, an ultrasharp tip with a radius of about 50 nm is still required to concentrate an electric field and eject atoms from a sample. In APT, researchers deduce structural information about a specimen from the trajectory of these atoms—which become ions—and from the points at which they strike the instrument’s detector. Flight times reveal ions’ mass-to-charge ratios and, hence, their chemical identity. Powerful computers then collect and crunch all these data to construct a 3-D rendering of a specimen.

To prepare the sharp samples, researchers have typically used electropolishing, a traditional electrochemical method. This technique was used to make nearly all of the specimens shown on the cover of this issue of C&EN.

Unlike electropolishing, a newer method does not require choosing suitable electrolytes and is amenable to preparing samples from a wider variety of materials. The technique combines focused ion beam (FIB) sputtering—atomic-scale sandblasting, typically with gallium ions—to sharpen the tip and scanning electron microscopy to observe and provide feedback on the microscopic sharpening process as it occurs. These so-called dual-beam FIB systems are now available commercially.

Another key advance, according to Thomas F. Kelly, a vice president with Philadelphia-based electronic instruments firm Ametek, is the integration into APT instruments of pulsed ultraviolet lasers, which assist in evaporating ions from the specimen tip. Cameca, an Ametek company located near Paris, is currently the only commercial manufacturer of APT instruments.

Earlier tomograph designs relied on rapid voltage pulsing of a small electrode positioned close to the sample. That design works well for metal specimens but is unsuitable for extracting ions from electrically insulating materials. Laser pulsing gets around that limitation by transferring thermal energy to the specimen tip, explains Kelly, whose APT company was acquired by Ametek in 2010 and folded into Cameca. The energy excites atoms in the tip and causes them to fly off as ions. The laser deposits very little energy per pulse, which spares the specimen from being heated too high and breaking.

That variant of APT analysis has been applied to a wide variety of sample types. Last year, for example, Oussama Moutanabbir of Montreal Polytechnic worked together with Seidman, Northwestern scientist Dieter Isheim, and others to probe the doping process in silicon nanowires. Doping semiconductors with select impurity atoms is an industrywide method for customizing a material’s electronic properties for various applications. Yet atomic-level control and understanding of doping processes remain elusive and difficult to probe via electron microscopy.

The team grew the nanowires with a vapor deposition method catalyzed by aluminum nanoparticles. From APT analysis, the researchers found that aluminum triggers a self-doping process, resulting in an unexpectedly high concentration of aluminum atoms evenly distributed throughout the wires (Nature 2013, DOI: 10.1038/nature11999). Because the presence of aluminum enhances silicon’s charge conductivity, self-doping might be a simple way of avoiding time-consuming doping of full-grown nanowires, a procedure commonly used in industry. The researchers developed a solute-trapping model to explain the findings, which they propose may lead to nanowires with tailored shapes and chemical compositions.

Metal nanoparticle catalysts can mediate nanowire growth. More often, however, the tiny materials are used to facilitate reactions in industrial processes. In these cases, scientists usually disperse the nanoparticles on an oxide material. Scientists would like to use APT to better understand the structure of these metal-insulator hybrids.

Arun Devaraj of Pacific Northwest National Laboratory (PNNL); François Vurpillot of the University of Rouen, in France; and coworkers recently analyzed samples of magnesium oxide-embedded gold nanoparticles with APT. This material exhibits surprisingly high catalytic activity for carbon monoxide oxidation, a key step in scrubbing automobile exhaust. By correlating tomography results with transmission electron microscopy (TEM) data, the team learned that metal-insulator combos can be trouble for APT. MgO evaporates preferentially, which degrades the method’s spatial resolution and the accuracy of metal nanoparticle composition analysis. The researchers determined that the errors can be corrected, however, via a theoretical treatment that accounts for differences in the components’ properties (J. Phys. Chem. Lett. 2014, DOI: 10.1021/jz500259c).

Some analytical techniques require fresh samples. APT isn’t particular about such things. In fact, the method is in some ways ideally suited to analyzing really old samples, ones dating to the dawn of the solar system and even earlier.

Because APT measures the mass-to-charge ratio of all ions impinging on a detector, the technique is a natural when it comes to analyzing isotope abundance—the key to radiometric dating. A team led by John W. Valley, a geoscientist at the University of Wisconsin, Madison, together with coworkers from Puerto Rico, Australia, and Canada, exploited the isotope analysis feature and APT’s subnanometer spatial resolution to date samples of Earth’s oldest known minerals. These zirconium silicates, or zircons, came from the Jack Hills, in Western Australia. Earlier studies concluded that those specimens are 4.4 billion years old. But because of radioactive decay processes that can cause atoms to diffuse in and out of zircon crystals and skew the dating results, that value is the subject of much debate.

The dating technique calls for quantifying uranium and lead isotopes, which must remain fixed in place inside a specimen for analysis to be a success. APT analysis of FIB-sharpened zircons recently revealed that lead isotopes are mobile enough to form clusters. According to Valley and coworkers, however, those changes occur on such a small scale that they do not affect dating results. As such, the team’s findings confidently confirm that the oldest known zircon on Earth is about 4.37 billion years old (Nat. Geosci. 2014, DOI: 10.1038/ngeo2075).

Philipp R. Heck, an assistant curator at Chicago’s Field Museum, together with a team of researchers from about 10 institutions, applied similar methods to determine the carbon-12 to carbon-13 ratio in meteoritic nanodiamonds. Samples, which the team embedded in platinum tips, came from the thoroughly studied Allende meteorite, a massive rock that predates the solar system and broke up over northern Mexico in 1969.

The team’s aim was to determine whether carbon’s isotope distribution in the nanodiamonds differs markedly from its distribution on Earth. Finding such a sample would suggest that the carbon was formed via a nucleosynthesis method different from the one that formed Earth’s carbon. Major differences in isotope abundances have been measured in other extraterrestrial material. So far, the nanodiamond study has not turned up obvious differences in carbon isotope distributions.

But it’s not “case closed” just yet. The analysis has proved difficult because it pushes APT’s detection limits. The nanodiamond study requires accurately counting and sorting very small numbers of atoms from approximately 3-nm-diameter particles and spotting a difference from the earthly ¹³C abundance, which is only 1.1%. The investigation has so far shown that APT is well suited to the cosmic science task, but larger data sets need to be generated (Meteorit. Planet. Sci. 2014, DOI: 10.1111/maps.12265).

Back on Earth, numerous organisms form mineralized tissues with unique properties and function through subtle interactions of organic matrices and minerals. As materials such as shells, bones, and teeth grow, the organic-inorganic interfaces that direct tissue growth become buried inside the tissue, making them largely inaccessible to imaging techniques. As such, researchers cannot fully determine the secret to these hybrid materials’ unique properties such as shape and strength, limiting their ability to mimic them in synthetic materials.

To test whether APT could help reveal the structure and composition of such hidden interfaces, Northwestern materials scientists Lyle M. Gordon and Derk Joester used FIB methods to prepare specimens of teeth from marine mollusks known as chitons. These critters’ teeth are remarkably hard and wear-resistant—capable of chewing through rock.

In 2011, the team showed that, indeed, APT can image hidden interfaces in these teeth with exceptional detail. The study revealed 3-D chemical maps of organic fibers 5–10 nm in diameter surrounded by nanocrystalline magnetite, Fe₃O₄. Unexpectedly, similar-looking organic fibers turned out to be chemically distinct. Some fibers hosted clusters of sodium ions—others magnesium. The researchers propose that the cations control mineral-matrix interactions, water retention in the tissues, and other factors that prevent the teeth from becoming brittle and affect their ability to accommodate large strains and dissipate energy (Nature, DOI: 10.1038/nature09686).

If mineralized mollusk tissue yields to APT analysis, hard vertebrate tissues may soon follow suit. That line of reasoning led Gordon and Joester to try forming needlelike specimens from nanocrystalline biological apatites—the mineral phase of vertebrate bone and tooth. As with other electrically insulating materials, it was not obvious at the outset that the researchers could prepare robust specimens or that the specimens’ nanoscale structural and chemical complexity could be scrutinized via APT.

Yet things turned out well. The team generated the first 3-D reconstruction of a roughly 10 million-atom dentin sample (ACS Nano 2012, DOI: 10.1021/nn3049957). The results show the fibrous nature of the collagen organic matrix in dentin, which is one of the components of teeth.

“We also found that the interface between the mineral and organic phases is nowhere near as sharp as transmission electron microscopy images imply,” Joe­ster says. TEM shows flat platelets of crystals. “Atom probe paints quite a different picture,” he stresses. “It’s going to be really interesting to find out why.”

Thus far, most biomaterials that have been subjected to APT analysis are the “hard” type, such as bone and tooth. A number of researchers, including PNNL’s Daniel E. Perea, are working to develop general procedures for analyzing soft biomaterials. One strategy Perea and others are pursuing involves freezing soft biomaterials by cooling them to cryogenic temperatures and maintaining them in that frozen state from sample preparation through the conclusion of APT analysis.

Field-ionization techniques used to be of interest mainly to metallurgists. But the past few years have turned up a dizzying array of materials capable of revealing their innermost structural and chemical secrets to APT. Classes of materials that once seemed out of bounds to the field’s early practitioners now are welcome in the lab.

Reflecting on those fast-paced changes, Northwestern’s Seidman says, “If someone asked me today, ‘Is it worthwhile to look at this material or that one?’ I would say, ‘Yes, you should try anything.’ ” Glad his high school chemistry teacher was proven wrong, he adds, “There’s no guarantee it will work or that you’ll get beautiful results. But it’s definitely worth a shot.”