SEM and AFM, a natural combination

Why to use a combined SEM - AFM? This combination offers several benefits, yielding to a better AFM as well as a better SEM, when comparing the combined system with two standalone systems. Actually, this combination is even somehow "natural", but more about that later.

At first now the advantages from the point of view of the AFM user:

Advantages from the point of view of AFM users

The AFM is a high-resolution-only microscope. When scanning, the AFM tip has to follow the surface in constant distance with nano meter precision. If it is not, in the best case it is too far away and one simply doesn't get all information from the surface, but in the worst case it is too close and both tip and sample are damaged.

From this results a maximum scan speed, which is that speed where the AFM tip just follows the surface without damaging the tip and the surface. This speed can be regarded as the "lowest resolution", as one will still see larger details on the surface but not as much as scanning with slower speed. This maximum speed is independend from the scan area size, i.e. magnification. With normal cantilevers, this speed is in the order of magnitude of 100 µm/s (see What determines the maximum scan speed of an AFM?).

When working with a light microscope, one commonly starts with a low magnification objective for easier focussing and for having an overview over the sample. Then one is using a higher magnification at the place of interest. If one works like that with an AFM and for example scans an area of 1 mm² with 500 x 500 pixels, which corresponds to a normal low magnified light microscope image, this takes more than 80 minutes with the speed mentioned above. Even more, if the sample is not absolutely clean and flat, there is a high probability that one has contaminated the tip with that long scan. If one finds now an interesting area in the image, one would like to zoom in further. But this makes only little sense with a contaminated tip, so, one is changing the tip and then starting a larger overview scan again because by changing the tip one commonly moves at least some micrometers on the sample and looses the interesting place. For mechanical setups, micrometer precision is quite good but for the high magnification of the AFM some micrometers are really a lot.

A large scan with an AFM takes very long and is bad for the tip. For an unknown sample, one is more likely working the opposite way, starting with a high magnification and lowering it after one has an idea about the sample roughness and contamination. Therefore, the AFM nearly always needs to be combined with a low magnification technique, to get an overview over the sample and find the places of interest. This is the reason why nearly all relevant AFM manufacturers deliver their AFMs together with an optical microscope for rough positioning. Unfortunately, the resolution of optical microscopes is strongly limited and some features on the sample surface like small topography changes etc. can not be seen at all with an optical microscope. This kind of features one can simply not find with a conventional AFM as long as they don't occur with high density all over the surface. With an optimized optical microscope with high numerical aperture one even compromises the AFM performance by reducing the stability of the whole system, but one can never close the resolution gap between the AFM and the optical rough positioning setup.

The electron microscope as a natural improvement of the rough positioning system

When one replaces the optical microscope with a scanning electron microscope (SEM), one can totally close the resolution gap, at least on a lateral scale. The SEM has the dramatic advantage that it works with low as well as with high magnification. With such a setup, one can find features that are simply impossible to find with another method. One can observe the sample condition, e.g. contamination etc., before contaminating and wasting the tip. One can even see the tip status itself and the tip interaction with the surface and does not unnecessarily work with used up tips which cause artifacts in the AFM image. This save a lot of time and money. Here a short list of the advantages:

• Exact positioning of the tip, location of features that can not be found with standalone AFMs
• Sample condition evaluation before the AFM scan and reduction of tip consumption
• Live tip evalution and observation of tip contamination during scanning
• Much faster and more efficient AFM work
• Possibility of doing combination measurements (e.g. influence of electron beam on surface potential)
• In-situ sample and tip preparation using a FIB: If the tip is dirty, just clean it!

From this list it is easy to see that an AFM within a SEM is much more powerful than a standalone system.

Advantages from the point of view of the SEM user

Not only the AFM user has advantages of putting the AFM into an SEM, also the SEM user benefits from an integrated AFM. What information does a secondary electron beam image contain? It is more or less a qualitative information of different material and charging behaviour. In many cases this is enough, but one does not obtain 3D height information and not at all with atomic resolution. To measure a surface profile, one needs to cut the sample or make a FIB hole.
But with an AFM one obtains on a button press a 3D image with atomically resolved height scale and calibrated X,Y,Z scales without desturbing the sample. This is the ideal addon to every SEM image. One can immediately see whether the bright spot on the surface is something sitting on top or a defect in the material. One can not obtain this information with any other method.

Another advantage: The AFM is an ideal sample manipulator: The tip approaches the surface by itself and standard cantilevers are sharp and more or less mass products. With an AFM tip one can easily mechanically manipulate the sample surface, compare stiffness and hardness, make electrical contact, measure temperature, and much more. And what does actually happen when one scans with an electron beam and measures at the same time the current through the AFM tip or vice versa scans with the AFM tip and reads out the SE detector, maybe with applied voltage on the tip? Here follows a short list of avantages for the SEM user:

• Calibrated 3D height information on button press
• Atomically resolved height information
• Ideal sample manipulator with tips as mass product and automatic sample approach
• Information about hardness, stiffness, temperature and more
• Combination measurements deliver information that cannot be observed with any other method

This shows as well, that a combined system is more powerful than the standalone systems.

Combination of SEM and AFM is natural?

Have you ever tried to find a place you identified in an electron microscope with an AFM? Very difficult thing. One has to use markers and many images, but this is never fast and when one has found the place with the AFM, the tip is dirty because of many tries. The opposite way is also difficult, one has only the optical image from the rough positioning system to identify the right place on the sample.

Why is it so difficult to switch between AFM and SEM and investigate the same position in both systems? Imagine, you have found something with few nm size in the SEM on a sample of normal CD size and want to find this again in the AFM. One could just think of scanning the whole CD with an AFM. How long would that take? Let's do a rough calculation...
We should scan with optimal resolution with the AFM, lets say 7 nm, which is the typical tip radius of a new AFM tip, to find something with nm size. Let's say we scan with 20 µm/s and do 5 µm x 5 µm images with 1024 x 1024 pixels, so that we end up with a pixel resolution just better than the 7 nm. If you measure with this scan speed, each image takes about 512 s, which is 8.5 min (when ignoring overscan range, TipGuard(TM) etc., and every line is scanned AFM typically forth and back). The scan of the whole CD (0.011 m²) then takes 2.2E11 s. This is about 7000 years. Who knows whether mankind will survive this scan? At least, the right place on the CD is found already after 3500 years on average. Generations of humans will eagerly await this moment and there will be a big party when this has taken place.

These extreme duration shows, how large magnifications we speak of, and that one should never start a long AFM scan because one will have died before the scan even really started at all. For an efficient investigation of the same place with extremely high resolution methods, like AFM and SEM, one simply needs a somehow integrated system. If you use markers etc. for this switching, finding the right place on a CD will not take 3500 years, but even 0.1 promill of that time is too long.

Besides the high resolution of SEM and AFM there is another similarity. Both methods actually only act on single points, one has to scan this point over the sample surface to get an image. Both methods therefore have similar image parameters: Scan speed, pixel distance and number of pixels. This defines the image format. So, what comes closer than using the same user interface software for both methods? We are doing it like that.

There are several similarities between AFM and SEM. Both methods provide important benefits for the other method, respectively. The combined SEM/AFM represents in fact a natural step ahead.

Back to BRR, the combined SEM AFM