Dreaming big, focusing small
Atomic resolution and further: Professor Stephen Pennycook has stretched high resolution imaging to its limits.
In March 2010, a team of one dozen microscopists and materials scientists used Z-contrast scanning transmission electron microscopy with aberration correction to resolve individual light atoms on a single-layer of boron nitride.
Brighter nitrogen atoms could easily be distinguished from darker boron atoms in hexagonal ring after hexagonal ring within the boron-nitrogen structure. Meanwhile three types of atomic substitution had also been resolved as well as the 0.1 angstrom in-plane distortions that ensued.
The team was led by world-class microscopist, Professor Stephen Pennycook, and included Nion co-founders Ondrej Krivanek and Niklas Dellby as well as leading researchers from Oak Ridge National Laboratory, Vanderbilt and Oxford Universities.
The instrument was a 100 kV Nion UltraSTEM operating at only 60 kV to, as Pennycook explains, 'avoid atom-displacement damage to the sample'.
And crucially, the images adorning the front cover of Nature unequivocally demonstrated the power of Z-contrast imaging - which distinguishes elements according to atomic number - with aberration correction.
Individual boron and nitrogen atoms are clearly distinguished by intensity in this Z-contrast STEM image from Oak Ridge National Laboratory. A hexagonal ring of the boron-nitrogen structure (as marked by the green circle) consists of three brighter nitrogen atoms and three darker boron atoms. The lower (b) image is corrected for distortion. [ORNL]
Yet, for Pennycook, these results are just one example of the many STEM imaging breakthroughs that have peppered his long-standing career.
Drawn to physics at a young age, Pennycook had always wanted to understand how things work rather than memorise endless chemical formulae.
So on completing his Masters degree in Natural Sciences at the University of Cambridge, he embarked on a materials physics-oriented PhD at the Cavendish Laboratory, using STEM to investigate cathodoluminescence in a range of materials.
The project came at a time when STEM was being used in biological sciences to map protein complexes and more, but had not yet gained momentum in materials science as diffraction contrast effects masked the Z-contrast during imaging.
"My supervisor basically offered me this project on cathodoluminescence and I thought it sounded intriguing so said, 'Okay, let's try it'," he says.
Working with a VG Microscopes HB5 STEM, Pennycook developed a cathodoluminescence detector and was able to generate high resolution cathodoluminescence images of dislocations in divalent oxides and diamond.
But at the same time, fellow student Michael Treacy, now Professor of Physics at Arizona State University, was also using STEM to study palladium and platinum catalysts.
Treacy wanted to develop a high-angle annular detector for STEM, to generate higher resolution and contrast images of the complex microstructure of his catalysts.
So together, he and Pennycook modified the cathodoluminescence detector to collect high-angle scattered electrons and image the catalysts with fewer diffraction effects.
As Pennycook puts it: "My cathodoluminescence detector turned out to do just the job, and this is what started the entire field of high-angular annular dark field STEM imaging."
By 1979, the researchers had obtained the first high angle annular dark field (HAADF) images of catalyst particles with much improved particle contrast with Pennycook then spending the next few years honing the method of Z-contrast imaging.
Z-contrast image of Si (a) and GaAs (b) dumbbells captured with a 300 kV VG Microscopes HB603U STEM [Adapted by Pennycook et al, Philos. Trans. R. Soc. A 354 (1996) 2619–2634]
Come 1982, the young researcher left Cavendish Labs to take up a staff position at Oak Ridge National Laboratory.
In his words, he 'wanted a proper salary', so he crossed the Atlantic intending to stay for around five years. His time at the Department of Energy (DoE) facility spanned more than thirty years.
In the beginning, ORNL didn't have a particularly high resolution electron microscope, but working in a small research team, Pennycook was able to use a high-angular annular detector with the STEM to get pretty impressive Z-contrast images of ion-implanted silicon samples.
"These samples turned out to be perfect for testing the idea of Z-contrast and investigating, for examples, how much contrast we could get from different dopants," he says.
At the same time, thoughts of atomic imaging were emerging. Other researchers were using high resolution electron microscopy to study atoms but still grappling with phase contrast interference.
Pennycook knew he could avoid these effects and achieve high atomic number contrast with HAADF STEM. And what's more, he reckoned if he equipped the instrument with an atomic-sized probe, atomic imaging would be within his reach.
"It was a sort of curious dream at the time," highlights Pennycook. "We asked ourselves what would happen if we had a beam of atomic dimensions and scanned it across particles, surely we would just be able to see atoms directly?"
Thankfully, the DoE stepped in and provided the researcher with a VG Microscopes 100 kV HB501UX.
The instrument was equipped with a high resolution pole piece and masks to exclude low-angle scattering from the annular detector, and come 1988, Pennycook had delivered incoherent, Z-contrast images of superconductor single crystals at atomic resolution.
STEM HAADF image of PbTe-5.5%Cu2Te showing copper interstitial arrays (top left) and copper interstitial clusters (top right). Enlarged images showing a high density of copper interstitial clusters (btm lef and right). [Pennycook]
Peter Nellist, now Professor of Materials at the University of Oxford, David Jesson, currently Professor of Nanotechnology at the University of Cardiff, and other up and coming researchers soon joined Pennycook's group which led to a long chain of theoretical papers explaining how incoherent, annular dark-field Z-contrast imaging worked.
Progress was rapid and the DoE soon provided Oak Ridge National Laboratory with a further STEM; a 300 kV VG Microscopes HB603U.
The instrument allowed the formation of a 0.126 nm diameter electron probe, which enabled Pennycook and colleagues to image the dumbbells in silicon and gallium arsenide as well as individual catalyst metal atoms on insulating surfaces.
"Nellist and I imaged single platinum atoms and published in Science," highlights Pennycook. "No-one had ever seen clusters this small in the catalyst world and this was quite pivotal."
"You know, it was Pete who really got our 300 kV machine working, and he often told us he had spent three years trying to resolve silicon dumbbells using ptychography in his PhD, yet did it in 20 minutes using our Z-contrast technique," he adds.
Working with Nigel Browning, now Professor of Physical Sciences and Engineering at the University of Liverpool and Matthew Chisholm, currently Electron Microscopy Group Leader at ORNL, Pennycook also went on to show that electron energy-loss spectra (EELS) could be used alongside annular darkfield imaging to obtain information about chemical composition at atomic resolution.
The research was published in Nature, in 1993, but much more was on its way.
Since the delivery of the 300 kV instrument, enthusiasm for using the original 100 kV at ORNL had dwindled, except for EELS analysis. However, change was afoot, namely in the shape of aberration correction, pioneered by Nion's Ondrej Krivanek.
In the early 2000s, Pennycook insisted that both ORNL's 300 kV and 100 kV STEMs had a corrector installed, and the results were instant.
"As soon as we put on the aberration correctors, the improvements were immediately obvious," says Pennycook. "We could see single atoms much, much better and with every generation of aberration correction we could see atoms more and more clearly."
"[The 100 kV microscope] now had the had the resolution of the uncorrected 300 kV STEM," he adds.
Together with Nellist, Krivanek, Niklas Dellby and many other colleagues, Pennycook quickly delivered result after result. In 2004, he demonstrated the first sub-angstrom imaging of a crystal lattice, as detailed in Science, as well as the first spectroscopic identification of a single atom within a bulk crystal.
'We don't yet have atomic-scale resolution in three dimensions, but we're getting very close.' Professor Stephen Pennycook
Alongside fellow researchers, Pennycook went on to image atomic configurations in catalysts, rare earth segregation in silicon nitride ceramics at subnanometre dimensions, atomic ordering at amorphous-crystal interfaces and much more.
Then in 2006, Pennycook and several ORNL colleagues were accused of research misconduct. An anonymous referee for a Nature Physics manuscript, on the use of STEM to measure charge transfer at oxide interfaces, had raised concerns over data manipulation and misrepresentation.
At the same time, an anonymous reviewer had also questioned the data used in Pennycook's seminal EELS paper, published in Nature, more than a decade ago.
"There were several papers that got attacked and it had been thirteen years since we published the one on EELS," says Pennycook.
"We had to put in such a lot of effort to try and dispel these claims that couldn't really be proved," he adds. "This was a huge distraction to say the least, and not just for me but for many other researchers in our group".
Two years later, a panel of independent external investigators unanimously concluded there was no evidence of research misconduct.
Pennycook and colleagues quickly moved on, and by 2010, the breakthrough results on atom-by-atom imaging of boron, carbon, nitrogen and oxygen were published in Nature.
Again working with Krivanek, Dellby, Chisholm and more, Pennycook had, for the first time, identified each atom in monolayer boron nitride, directly from intensity in an annular darkfield image taken using Z-contrast STEM with aberration correction.
The researchers had used a Nion UltraSTEM with a low accelerating voltage, heralding a new trend in atomic resolution imaging. Indeed, Pennycook and other researchers went onto use the same pioneering low voltage method to image graphene, nanocrystals, complex oxide thin films and much more.
"We got up to fifth order correction, before I left ORNL... aberration correction [with Z-contrast imaging] was such a big part of that decade," says Pennycook.
Following his retirement from ORNL, Pennycook took joined the National University of Singapore, where he remains today.
"The University wanted to start a new microscopy centre," highlights the researcher. "It had lots of advanced research and applications in many fields, including graphene and oxide electronics, but no advanced microscopes."
So the University brought Pennycook to Singapore, provided a fifth order aberration corrected STEM from JEOL and he set to work looking at atomic-scale defects in 2D materials, switching in interface-based devices and more.
"JEOL could provide the instrument very quickly and it fits in with our research very well," says Pennycook. "The probe performing angle is about a factor of two larger than previous generation machines which means that the resolution improves not just in the X and Y directions, but also in the beam, or Z, direction, giving nanoscale depth resolution."
In 2013, Pennycook worked with ORNL colleagues to directly atomically-resolve a single Si6 cluster trapped in a graphene nanopore. [ORNL]
Indeed Pennycook and colleagues have been looking at nanoscale heterostructures for next generation electrocatalysts, critical for hydrogen production, the structural origin of piezoelectric performance in lead-free ceramics and much more.
At the same time, the researcher and ORNL colleagues have also been using STEM to 'sculpt' crystalline oxides, at the atomic level.
"With a 40 kV beam, you can peel off, for example, the selenium from a molybdenum diselenide monolayer and make a metal monolayer," says Pennycook.
But for Pennycook, the real excitement lies in 3D atomic imaging and he reckons it could be realised very soon. "We don't yet have atomic-scale resolution in three dimensions, but we're getting very close," he says. "I don't think we will get there with the current generation of aberration corrector but if we can expand the probe angle a little more then yes, we could get to atomic resolution."
Right now, his team is also enhancing resolution with mathematical methods, such as phasing and deconvolution, as well as combining simulations.
"We might tilt the sample so the atoms are not right under each other, and we could also slice through materials; if we're looking at a nanometre-thick section, who knows what might we see?," he says. "Don't be afraid to dream, as sometimes, dreams can come true."