I have been using a Zeiss SEM for imaging 10-50nm diameter gold particles. To minimize charging, we use low energies (1 kV), and have imaged the particles in various detectors (ie in-lens and SE2).
Our smaller particles often have a halo around them. My question is, what is the theoretical vs. practical resolution limits on a 1kV beam? I know that in visible optics, microscopes are at their theoretical limits, but in IR microscopy, often we can't get as close to the theoretical limits due to limitations in optical components (although I'm sure there are groups who are close by now).
What would you guys estimate is the smallest length scale that I could pragmatically resolve with a 1kV beam? Would the SEM detector type drastically affect this in any way?
I'm a novice obviously!
Dear Adam,
in my opinion the theoretical resolution for 1kV electrons is absolutely not reached at the moment. There are great improvements going on in the last years that improve the maximal resolution at low energies. For example, energy monochromators, beam deceleration or immersion optics, in which the sample stays inside the magnetic field, have proven to enhance the resolution to nearly 1nm.
Since the research in this field is going on in a very fast and dynamic way, I would expect further improvements in the future and I hope for a resolution much below 1nm.
The detectors play only a little role in this resolution battle, I would say. The fine focusing of the beam is the main challenge.
Dear Adam,
in my opinion the theoretical resolution for 1kV electrons is absolutely not reached at the moment. There are great improvements going on in the last years that improve the maximal resolution at low energies. For example, energy monochromators, beam deceleration or immersion optics, in which the sample stays inside the magnetic field, have proven to enhance the resolution to nearly 1nm.
Since the research in this field is going on in a very fast and dynamic way, I would expect further improvements in the future and I hope for a resolution much below 1nm.
The detectors play only a little role in this resolution battle, I would say. The fine focusing of the beam is the main challenge.
Thanks!
Do you think the lower limit is determined by the DeBroglie wavelength of the electron, or another criteria (like the Raleigh Criteria)? If the focusing is the problem as you point out, it would be the Raleigh criteria, no?
I found some nice explanations here: http://www.sci.sdsu.edu/emfacility/555class/class1.html
Following this, the electron wavelength is approx. 1.23/sqrt(acc.volt) which leads to 0.04nm for 1kV electrons. In optimal conditions (like they are already reached in optical microscopy) the resolution is half of the wavelength, --> 0.02nm. So there is a lot of room for improvement in the next years! :)
The focusing is the main problem, because you don't have perfect lenses for electrons. Also the charge plays a role, especially at low acceleration voltages. The electrons repel each other and thus the beam widens up. This is the most important point where recent developments like the beam deceleration or the immersion optics are effective.
An aspect that is often overlooked in the quest for ever higher resolution is a statistical analysis of the data observed. We recently published a study of graphene where we obtained sub-picometer structural parameters form a 50 picometer resolved TEM image:
Nanoscale 2013 march : Article: Sub-picometer structural information of graphene hidden in a 50 pm resolved image.
Steffen B Petersen, Gnana Prakash Gajula, Maria Teresa Neves-Petersen
Adam,
the resolution in SEM is primarily limited by the excape volume of the signal careers. Thus elemental specific imaging based of X-ray detection is limited by hundreds of nms, imaging in backscatered electrons - by tens of nms and in secondry electrons is energy and material dependent.
This is why a detector in SEM has a substantial impact in resolution: if you compare the images obtained by ETD detector and in-lens (Zeiss) /in-column (FEI) detectors you will see that the later are substantially sharper. This is because in-lens detectors can filter only secondaries with lower energies having smaller excape depth. This is also a reason for using as low HT as possible, especially for light materials like carbon. Possible means that at very low energies you will start getting all these problems mentioned by Bjorn.
Quality of the optics is important of course and is limited currently mostly by chromatic aberration. Gun monocromation (or cold FEG) is a feature of all top instruments nowadays.
More advanced customized instruments have even a corrector of chromatic aberration, and I have seen a publication showing atomic resolution in SE signal.
Commercially one can get an instrument (just signed off one of them in my lab last week) showing 0.6nm at 15kV and 0.8nm at 1kV in SE.
However in practicall operation the resolution is limited by contamination - below 0.5 nm already a monolayer on the surface starts to hide structural detailes, and most of top level instruments are not UHV tools.
Steffen,
The question of statistical treatment is a keystone of experimental physics in general and of microscopy in particular. It was never overlooked and there is a massive lot of literature on this topic.
In your publication you unfortunately misinterpret increase in precision due to averaging of measurements as increase of resolution - these two are completely different things. If you think of Gaussian distribution, the width will correspond to resolution and position uncertainty - to precision.
And this is a kind of schoolbook fact, that making multiple measurements will increase a precision, and in image processing there are much more efficient and accurate methods to obtain averages as compared to what you proposed in the paper.
Unfortunately this is not the only drawback of above publication – you have used our image and without knowledge of the topic has dramatically misinterpreted imaging artifacts as a revolutionary new structural feature of graphene, which supposedly was overlooked by thousands of highly experienced people working in the field, just because they did not know how to average data correctly. I really wonder how this submission passed through the referees.
Adam, for gold particles you can try higher voltage, it can give better resolution. Also, contamination could be a problem; you can try to focus first and then move to the fresh field of view and immediately take a picture. Good luck.
In STEM mode the limitations in excitation volumne of SE and Bck electrons can be overcome. But for low voltages like 1kV most sampes are not electron transparent anymore and no signal can be obtained below the specimen.
Nevertheless there is proof that on silicon in STEM and Bcksc electrons an atomic silicon lattice in 110 direction can be observed. The resolution is in the 0.21-0.31nm range depending on your definition of resolution.
Resolution definition is hard to define on gold particles, since you want to measure a distance between two objects to be resolved with a Raleigh criterium, which is difficult to do no nanoparticels and easy on peridoic objects
Cheer
The theoretical limit has obviously not been reached if all factors are taken into account. But the practical limit of the best SEM's in the world is 2-5 nm, even with the highest possible energies you may use. This limit is not due to the wavelength of the electron, but due to the geometrical factors related to the focusing of the beam. Using TEM, you can get higher resolution for the same electron energy used in SEM.
Why don't you try AFMs? These are cheaper than SEMs and has better resolution.
- Badges
- Science method