Much needs to be mastered in order to produce a fully functional electron microscope. Developing software is one of them. TESCAN has been a pioneer ever since its fledgling days in using computers to control microscopes. But our devices have come a long way since then, as software has taken on much more importance, as in other fields.
We sat down with Vojtěch Filip, head of SEM/FIB software development to ask how does software work in our microscopes, and what are some of the emerging trends and challenges.
You joined TESCAN more than 20 years ago, what brought you here?
It was actually a coincidence, I graduated from the Faculty of Electrical Engineering and Computer Science, and following my military service, I started working for an antivirus company based in Brno. The work was pure software development, something I didn’t really enjoy. I heard of the TESCAN vacancy through a former classmate. I was very interested in connecting software with hardware, so I joined up in 2000.
What did the SW development department look like back then? And can it be compared to the present?
There were four of us then, now there are over 30 of us here in Brno alone, other colleagues working in France (ORSAY), Belgium (former XRE), USA (TESCAN Tempe) and other autonomous development teams can also be found in our subsidiaries (such as TESCAN 3DIM). That perhaps give a good idea of how long we’ve come. I am in charge of software development for FIB-SEM microscopes, we also have an Embedded Software Development Department here in Brno, which is in charge of developing single-chip processors, that are hidden in the electronics and the customer does not see them anywhere on the screen. And TEM Software Development Department which is led by Ondřej Beňuš.
Since the very beginning, TESCAN has designed microscopes to be fully software-driven. It sounds obvious now, but analog devices were still very common even in the late 1990s. Twenty years ago it was very much about controlling imaging through the software, whereas today software plays a much bigger role throughout the electron microscope.
So what is the biggest challenge from a software perspective in electron microscopy today?
The biggest challenge is to simplify device control even as those devices become more and more intricate and complicated. We used to produce relatively simple microscopes (seen from today’s standpoint) without FEG sources, with only one two detectors, and we did not produce FIBs. Today, thanks to software, microscopic analyses can be done more accurately, and with modern user interfaces they have become easier to operate. Last but not least, much of this work can now be done completely automatically. Take for example, TESCAN TIMA: the entire sample analysis process is practically automatic, just load the sample into the microscope and hit “start”. Those samples take all day to be analyzed, but the user (researcher) can do something else and then just go back to retrieve the finished dataset. All thanks to software.
So you think the automation and simplification of service are trends that will continue?
Yes, it also has to do with how electron microscopes themselves are changing. The are now multimodal devices that, in addition to imaging, have plenty of other features that are useful for our customers: one can determine chemical composition using EDS, crystalline lattices using the EBSD detector, or use a mass spectrometer to determine the mass of the particles. Moreover, thanks to the FIB, we can now look below the sample surface, analyze molecules using the Raman spectrometry, etc. We are now able to put the sample through a number of techniques or any combination of them. So it’s our job to drive the whole thing software-wise, including integrating third-party detectors and equipment. Individual components must communicate with each other perfectly and function smoothly as a single unit.
So is it necessary to have powerful computer hardware to operate the microscope?
I wouldn’t say that; as a matter of fact, we can still get by with regular desktop computers. Of course, we also adapt the configuration to customers’ wishes and the intended use, but from a hardware perspective, it’s nothing that cannot be easily purchased. It is on the software development side where it becomes more interesting. The microscope software contains 5 to 10 million lines of code. That’s not exactly normal even for pure IT companies. If I have to put it in layman’s terms, I would compare the complexity of our software to the entire MS Office package. I expect that the volume of code there is similar, but no special hardware is required.
Since you mentioned Microsoft, how long do you keep the microscope software updated? Do you also issue regular updates?
It is quite similar, so I’d say that a 10 or 12-year-old electron microscope might still have up-to-date software. We certainly monitor compatibility with Windows, even for such old microscopes we issue an update pack every six months or so. Of course, our priority is to develop software for the current microscope series, where updates are issued every two months or so. It doesn’t work automatically because our customers don’t usually have microscopes online, but updates are installed by service personnel.
So developers in your department must have some idea of how electron microscopes work?
Basic awareness is certainly an advantage, but given both the complex devices involved and the integration itself, we take the help of colleagues from the Apps Development department. They usually sit in front of the microscopes, and focus on linking the microscope and the software, knowing the application workflow, talking to customers about how samples could be better processed from a software perspective, cooperating with institutions such as CEITEC and other scientific institutes abroad, and looking for new innovative analytical methods for obtaining data from samples.
So innovation and improvement requirements usually come from them?
Literally anyone can come up with a good idea at TESCAN. Fortunately, we are not yet such a big and rigid corporation that we need to have dedicated processes to do so. But we’ve created several channels through which each employee can put forward their ideas for technical solutions. We try to directly motivate people with financial rewards, publication opportunities or even by patenting the solution with their names as authors. This is how we would like to compete with the numerous regular IT companies that are now proliferating all around us. Our advantage is in the agility of development, and meaningful work with a deep connection to a real machine. If one has an idea, it can be quickly tested to get feedback from the lab just one floor below. We don’t archive it somewhere, but the results are extremely fast: you write the software one day, and in two months you have it in our microscopes all around the world.