It is extremely rare for developments in instrumentation to deliver an order of magnitude improvement in resolution or sensitivity for a particular analytical technique, while maintaining existing performance in other areas. A study by Krivanek et al.1 on page 209 of this issue, which focuses on high-energy-resolution electron-energy-loss spectroscopy (EELS) in a scanning transmission electron microscope (STEM), achieves just that. The results promise a new territory for experimental investigation of both fundamental physics and the structure of advanced materials at the atomic scale.

The work is a true collaboration between a small manufacturer of electron microscopes and several academics, including teams from two US universities that have recently installed this STEM/EELS instrumentation. It represents a good example of a research community working together to drive forward a particular technique. The paper describes the practical implementation of a design2,3 for an electron monochromator — a device that produces an electron beam that has a narrow electron-energy distribution. The electron monochromator acts on the high-energy electron beam of a STEM, which is corrected for aberrations in the lens that focuses the electron beam into a small probe of atomic dimensions4.

In a STEM, a high-energy electron probe is transmitted through a thin specimen, undergoing energy losses as a result of scattering from atoms in the sample. The probe is scanned over the specimen to form images of it as a function of probe position. If an electron spectrometer is added to a STEM, a spectrum of the transmitted electron intensity versus energy loss — the EELS data — is also obtained. Krivanek and colleagues show that, with the inclusion of appropriate energy-stabilization schemes, the monochromator design that they have implemented can achieve an electron-energy resolution for EELS in a STEM of around 10 millielectronvolts (Fig. 1), while retaining a sufficiently small probe diameter with enough electron current to maintain atomic-resolution imaging and analysis.

Figure 1: Spatially resolved vibrational spectroscopy.
figure 1

In a scanning transmission electron microscope corrected for aberration effects, a beam of high-energy electrons is focused onto a thin sample, forming a probe of atomic dimensions. The electron beam transmitted through the sample, which in this case comprises different types of atom and bond length, contains electrons that have experienced energy losses owing to scattering events with atoms in the sample. These electrons can be dispersed in terms of their energy using a magnetic prism spectrometer (not shown), enabling a spectrum of scattered electron intensity versus energy loss to be obtained. Krivanek et al.1 show that, if the incident electron beam goes through an energy monochromator (not shown), then the energy spread associated with a beam that undergoes zero energy loss can be improved from 250 millielectronvolts to 9 meV. This improvement allows the detection of spectroscopic signatures of spatially localized chemical-bond vibrations (phonons). The intensity scale of the phonon signal is magnified by a factor of 300 to show the phonon peak (blue arrow). (Plot adapted from ref. 1.)

Full size image

With an atomic-sized probe, STEM imaging using electrons scattered from atomic nuclei through high angles can already generate directly interpretable maps of the projected atomic structure of thin crystalline samples. In most materials, however, these images tend to be dominated by scattering from the heavier elements in the samples, making precise location of light elements difficult. EELS is an established technique in the STEM for studying electronic excitations in thin samples that are induced by the high-energy electron beam5. Such spectra can already provide maps of the distribution in elemental composition and chemical bonding at atomic resolution in two and, more recently, three dimensions.

The electron-energy resolution for EELS demonstrated by Krivanek et al. now enables the identification of energy losses associated with lattice vibrations in solids, in particular those related to longitudinal optical phonons6. These are out-of-phase lattice vibrations in the plane of the thin sample, arising when neighbouring atoms in the lattice have different charge or mass.

Until now, lattice vibrations were something electron microscopists have had to worry about only in terms of the sample damage that they induce, or when matching experimental images to simulations7. However, the main point of Krivanek and colleagues' work is that optical phonons are key signatures of chemical bonds, particularly those involving light elements such as hydrogen, as is well established by the techniques of infrared and Raman (optical) spectroscopy. The implication is, therefore, that STEM–EELS may provide a route for the direct mapping of chemical bonding, including that associated with light elements, at near-atomic resolution.

This achievement would present tremendous benefits in a number of highly topical areas of research into new advanced materials and devices. The improvements in overall energy resolution of EELS will undoubtedly aid the study of the local spatial variation of energy bandgaps in semiconducting structures, and the identification of localized collective oscillation of electrons in 'plasmonic' structures for light capture. The ability to detect and map light elements, including hydrogen, could extend the existing capability of analytical electron microscopy to the study of organic materials such as polymers and pharmaceuticals, as well as energy-storage materials — if issues associated with electron-beam-induced damage can be addressed. Directly measuring phonons could potentially help to identify chemical reactions involving the surfaces of nanoscale heterogeneous catalyst particles, and could aid the investigation of the transmission of lattice vibrations across micro- and nanostructural features, such as interfaces and defects in thermal and optical materials. As with the emergence of any new technique, many additional research areas may ultimately prove to be most fruitful.

Existing theory suggests that electrons that have undergone phonon scattering would be scattered through large angles and the resulting phonon signal would be spatially highly delocalized, preventing atomic-resolution analysis. However, Krivanek and colleagues present some initial findings which, together with recent theoretical predictions8, suggest that under appropriate conditions the phonon signal may be sufficiently localized for the study of vibrations at a spatial resolution better than that achieved by scanning probe tip-enhanced vibrational spectroscopies9. The authors observed an exponential delocalization of the phonon signal as an electron probe is moved away from the surface of a sample and into the surrounding vacuum. However, there seems to be a more localized component of the signal that peaks in intensity close to the surface itself, and the researchers discuss a possible experimental geometry for signal collection that would enhance this more localized contribution. Furthermore, if the probe is inside the sample, it seems that the delocalization could be screened at the interface between two materials with different electrical properties.

The authors also demonstrate a method for remotely exciting such phonons at a surface using the inherent delocalization of the signal; here, the beam is located in the vacuum close to the edge of a sample, potentially helping to mitigate electron-beam-induced damage of radiation-sensitive samples. These investigations of the spatial resolution of the phonon signal represent a clear example of experiment leading theory in terms of the interpretation of the results, and is indicative of the new experimental landscape that this development in instrumentation unfolds. Undoubtedly, many more exciting experiments with this technology will follow, aided by the delivery, later in 2014, of a third-generation instrument to a shared user facility: the Engineering and Physical Sciences Research Council National Facility for Aberration-Corrected STEM, or SuperSTEM10, in Daresbury, UK. I look forward to the community charting this new frontier of research.