Ondrej Krivanek: The physicist who changed microscopy
Image: Ondrej Krivanek's imaging breakthroughs have revolutionised electron microscopy.
From very early on, Krivanek was drawn to physics and mathematics. His father, a chemical engineer, had a passion for the chemistry of colour photography, and as Krivanek says, also had ‘near-photographic memory’.
"I didn't have this kind of memory and so analysed things instead," he tells Microscopy and Analysis. "In maths and physics, you can derive nearly everything from first principles and understanding is much more important than photographic memory. So maths and physics were definitely for me. I also liked practical applications, and that’s why physics won out."
The young Krivanek also loved constructing model aeroplanes, a hobby that he believes fanned the flames of the many instrumentation projects he took on later.
And growing up in Prague in 1960s Czechoslovakia, he appreciated the educational bias upheld by an otherwise difficult socialist system at the time.
Early days: A young Ondrej Krivanek camping with his sister Katia in Poland.
Come 1968, and in his senior high school year, Krivanek was invited to join the national team representing his country in the 2nd International Physics Olympiad.
His team came joint second with Hungary and East Germany, and importantly for the young student, it was fun.
"We got to work on problems that were too challenging for regular school textbooks,” he says. "I wasn't the most diligent student in regular classes, but when challenged by tougher problems, I got into them and wanted to know what the solutions were."
Krivanek went on to pass the entrance exam for a Physics degree at Prague’s Charles University, with classes starting in October.
The Summer before, Krivanek secured Visas to travel to France and the UK; as he boarded his train to Paris, his father warned him that if the Soviets were to invade whilst he was away, he should stay in the West.
It was the year of the Prague Spring, invasion ensued and Krivanek stayed in the UK, accepting a place at the University of Leeds as his family emigrated to Switzerland.
Krivanek loved Leeds but come the early 1970s, he headed to the Cavendish Lab at the University of Cambridge to study amorphous semiconductors by electron microscopy for a PhD.
His supervisor was physicist Archie Howie, renowned for his interpretation of TEM images of crystals, and at the time, many researchers were looking to use these relatively obscure materials in solar cells, semiconductor memory and other applications.
"Studying the structure of amorphous materials by direct imaging was over-ambitious back then," highlights Krivanek. "We only had around three Ångstroms resolution in X and Y, and much worse in Z. That meant that our ‘information channel’ had too little bandwidth to determine the positions of all the atoms.”
Still, Krivanek managed to detect ordered regions as small as 15 Å in so-called amorphous carbon, publishing his results with Howie and Philip Gaskell in "Seeing order in 'amorphous' materials" in Nature, in 1976.
"I learnt a little about the materials that I was studying and a lot about electron microscopy," says Krivanek. "Back then, electron microscopes were large, slightly mysterious machines and you could see things down at the atomic level; it was fascinating."
Key figure from Krivanek's 1977 Ge grain boundary paper. As he explains, the experimental micrograph was so clear, he could simulate it by building a model and then taking an out-of-focus picture of the model, (see 3c).
Post-PhD, Krivanek spent three months at Kyoto University, working with the then world's highest spatial resolution microscope; a 500 kV device with a low spherical aberration objective lens and stable sample stage.
In the researcher's words: “This was difficult to operate and you needed a bit of luck to get good results from it.”
Luck or not, Krivanek also developed what can now be described as one of the first tuning algorithms, allowing him to produce spectacular images from a trying instrument.
And come 1977, together with Japanese colleagues Seiji Isoda and Keinosuke Kobayashi, he published atomic resolution images of complex defects at a grain boundary in crystalline germanium in Philosophical Magazine.
As Krivanek says: "This work was an early harbinger of new capabilities made possible by high resolution electron microscopy, and really opened up a lot of people's eyes."
And open eyes it did. Following Kyoto, and a three month travel stint across Asia, Krivanek headed to Bell Labs, New Jersey, US, to take a look at the many interesting materials at the laboratory.
Bell Labs had yet to embrace high resolution electron microscopy, and Krivanek - now establishing his reputation as a world-class high resolution electron microscopist - used to travel to Cornell to image semiconductor samples.
He obtained the first atomic resolution images of silicon oxide interfaces.
"This type of interface is the workhorse of field effect transistors, but back then nobody knew what it looked like," he says. "Go to Intel today and you'll see many electron microscopes doing quality control for devices day in day out, yet in the late 1970s, nobody had any idea that electron microscopes could be useful for this."
From Bell Labs, Krivanek moved to the University of California, Berkeley. By this time, high resolution electron microscopy was maturing, and improvements were increasingly incremental and 'were more for the microscope manufacturers to do'.
However, a 1978 Cornell workshop on the blossoming field of analytical electron microscopy sparked Krivanek’s interest in electron energy loss spectroscopy, and given the lack of commercial instruments at the time, he decided to build his own.
"With EELS, I realised you could do much more than image structures, you could explore the composition, bonding, and electronic properties," he says. "Building a spectrometer was risky as I hadn't built any major instruments before, but I used some of the practical skills I had learnt from building model planes and it worked out well."
With $10k in university funds, and alongside EELS data software from fellow physicist Peter Rez, Krivanek built what would become known as 'the Berkeley spectrometer', kick-starting his career as an instrument designer.
"In the beginning, we had five people at Nion, but we thought move over FEI and JEOL, we'll make the world's best electron microscope." Ondrej Krivanek
Soon afterwards, he joined forces with Peter Swann, ex-Imperial College London Professor and co-founder and president of Gatan, and the pair developed a higher performing, more user-friendly instrument for commercial applications.
"We were on the same wavelength and for the first time, I was working with someone who was very good at designing things," says Krivanek. "Peter was a metallurgist who had taught himself design; he was also a perfectionist and did some amazing things."
It was now the early 1980s and as well as consulting for Gatan, Krivanek had taken the post of assistant professor at Arizona State University.
At that time, ASU was a great magnet for researchers, having recruited John Cowley who then attracted John Spence, Sumio Iijima, and many others, shaking up the world of high resolution electron microscopy.
Given his interest in EELS, Krivanek persuaded Gatan to donate a serial spectrometer he co-designed to ASU. The pioneering results that followed included work on electron channeling, EELS of surfaces, and EELS-CL coincidence.
Together with physicist, Channing Ahn, Krivanek also produced the EELS Atlas, still a standard EELS reference 35 years later. And mindful of how his interest in EELS was sparked by attending the Cornell workshop a few years earlier, he enthusiastically threw himself into organizing research workshops for schools.
"You always need a fresh perspective and at these events you can chalk up great collaborations for the future," he says. "I found this wonderful."
Nion founders: Ondrej Krivanek and Niklas Dellby with the company’s first employee, George Corbin.
But come 1985, Krivanek's Gatan colleague, Swann, had set up an R&D centre in the San Francisco Bay Area, so Krivanek moved back to California to help establish Gatan as a pioneering force in electron microscopy.
He was given a large, empty lab and put together a design team that developed several revolutionary products over the next few years.
His first instrument, a parallel-detection EELS spectrometer, improved detection efficiency by several orders of magnitude and could take the same quality spectra as a conventional serial spectrometer in 0.1 seconds rather than 100 seconds.
"Parallel detection made EELS into a truly useful technique, and from there we continued to develop one major product a year for about six years," says Krivanek. “Our imaging filter allowed researchers to produce elemental maps quickly and slow-scan CCD cameras took us down the path of replacing photographic film."
And as Krivanek also highlights, the high voltage imaging filter allowed EELS to be carried out in 1 MeV microscopes, DigitalMicrograph software became the standard for today's TEM software, and Digiscan showed how several STEM signals could be digitized simultaneously and helped to introduce practical spectrum-imaging.
Aberration Holy Grail
As part of the imaging filter project, Krivanek had successfully tackled strong second order aberrations and distortions.
He now knew that complete third order aberration correction was a classic problem in electron microcopy, known as “spherical aberration correction”, and wanted to work on it.
Many had looked at the problem, but nobody had produced a practical system that improved the microscope’s resolution.
Yet, Krivanek felt, given the right approach, correcting the spherical aberration of a scanning transmission electron microscope (STEM) would be straightforward and lead to spectacular improvements in resolution and analytical sensitivity.
"From my perspective it was simple," he says. "We had a second-order aberration-corrected spectrometer/imaging filter, and third-order correction did not seem much more difficult."
Krivanek was alone in his confidence to solve the problem, and was warned by some colleagues that pursuit of aberration correction would 'bury his career'.
Still, in 1994, he and colleagues, Mick Brown and Andrew Bleloch, secured Royal Society funding to develop a STEM aberration corrector at the Cavendish Laboratory in Cambridge. Niklas Dellby also joined the project.
As well as spherical aberrations, the researchers had to tackle 'parasitic aberrations'.
And unlike previous attempts, most notably by Albert Crewe in the US, Krivanek and Dellby did not think that eliminating the parasitic aberrations by highly precise construction was a practical proposition.
Instead, they incorporated additional optical elements to fix the parasitic aberrations, and developed auto-tuning software that would then measure all of the aberrations and fix them.
"We decided that the right approach was to admit that we could only build an imperfect corrector, and then use a computer to measure the [remaining] imperfections and fix them," says Krivanek.
And so the Cambridge STEM aberration corrector, as it became known, was delivered in 1997.
Krivanek’s time at the Cavendish was now up, and his family wanted to return to the US, so he took the post of Research Professor at the University of Washington in Seattle.
At the same time, he and Niklas Delby launched Nion, with the fledgling company focusing on making a second-generation STEM aberration corrector for a more modern STEM.
The development was largely funded by the IBM TJ Watson Research Center, whose Philip Batson ordered a second-generation corrector from Nion before its design was even started.
Open software collaboration: Participants in a workshop on Nion Swift open source software organized in Bad Mitterndorf, Austria.
When Nion delivered and installed the corrector into IBM's VG STEM in June of 2000, it was the first commercially delivered spherical aberration corrector in the world.
The microscope achieved 1.4 Å resolution during tests at Nion, and soon progressed to sub-Å resolution in Batson’s capable hands.
Daresbury SuperSTEM and Oak Ridge National Laboratories followed suit, with ORNL ordering two correctors that reached 0.78 Å direct resolution.
"We achieved about a factor of two improvement in resolution, which really woke people up," recalls Krivanek. "We had deep sub-Angstrom resolution and there was no way you could get that without aberration correction."
Not content with building aberration correctors, Krivanek and Dellby upped their game and set out to construct an entire electron microscope, with partial support from the US National Science Foundation.
The design was first built on top of 'VG's extra-bright cold field emission electron gun', but was vastly different from existing instruments with key innovations including high performance quadruple/octupole C3/C5 corrector, ultra-high vacuum at the sample, and a rotationally symmetric sample stage that minimised drift.
Three years later, Nion added its own field emission gun in a development partly funded by the US Department of Energy.
This increased the operating voltage to 200 kV, and improved on VG’s gun performance in other important aspects.
"Looking back, we really didn't lack ambition, did we?" laughs Krivanek. "In the beginning, we had five people at Nion, but we thought move over FEI and JEOL, we'll make the world's best electron microscope."
Cornell University and Daresbury SuperSTEM laboratory quickly adopted the new Nion UltraSTEM, while CNRS Orsay, France, ordered the first 200 kV version of the microscope.
With STEM aberration correction firmly established, Krivanek, as always, went on to pursue new directions.
It was 2008, and the energy resolution of EELS was coming in at around 100 meV for monochromated instruments, but the Nion founder had designed a new monochromator concept that he reckoned could reach 10 meV.
As in the past, his peers had their doubts, but thanks to his strong reputation in instrument design, Arizona State University placed an order, and Nion delivered.
In 2014, the ASU instrument enabled the first demonstration of vibrational/phonon spectroscopy in an electron microscope.
And today, attainable energy resolution is down to 5 meV, largely thanks to a new EELS spectrometer that Nion developed specifically for phonon spectroscopy.
Further instruments have been installed in the US, Europe and China, and as Krivanek states: "I think that with this development, Nion is revolutionising the world of EELS."
a) ADF STEM image of ice condensed onto an h-BN flake and onto lacey carbon. The e- beam was positioned at the arrow, where it drilled a hole through the ice. The circled region was then probed without further radiation damage by aloof EELS. b) Vibrational EEL spectrum of BN and ice. Nion monochromated STEM with side entry stage and liquid N2 holder. 100 kV, EELS spectrum acquisition time = 10 s. Courtesy Nion and CHROMATEM project, CNRS France.
Today, Nion employs more than 25 staff, and is home to a microscope R&D centre and an assembly line. Krivanek remains involved in design projects and is very excited about exploring the world of phonons using high resolution electron microscopy.
He is also investigating atomic resolution imaging at low temperatures, and says: "Typical low temperature stages shake in the breeze. We have developed, in collaboration with Henny Zandbergen in Holland, a liquid nitrogen holder that has reached one angstrom resolution at liquid nitrogen temperature, and very low drift rates.”
He expects that software will be crucial for future microscopists.
Highlighting how young researchers will need to use modern software such as Python to program microscopes for more complex experiments and data analysis, he believes such software should be open-source and freely shared.
"With sub-Ångstrom probes being common nowadays, the physical capabilities of electron microscopes are not going to evolve as fast as in the past," adds Krivanek. "The flexibility and power of microscopy is going to lie in what you can do with these probes, which means that smart and flexible software is going to be key.