Here is the cropped 1um X 1um image:

What additional information are you trying to get with the AFM over the information from the SEM? The advantage of AFM is that you can get height information, if you are interested in that. If you are more interested in shape and distribution, it might be preferable to use an image analyser on your SEM shots.

On the face of it, this could be a good sample for AFM. How possible it is, and how good the AFM image is will depend on a range of factors - how mechanically stable is the sample (you could be left with a swept hole in your particle distribution if they are very loosely adherent, alternatively you could pick up a particle or two and end up imaging with them) what's the height, width and spacing of your particles etc. What size and shape is the AFM tip, as the shape of the 'particles' in your image could be affected by this. Is there any material on the surface, such as solvent/dispersant that isn't visible on the SEM, but would show on AFM?

Generally an actual AFM scan takes just a few minutes, but it is the setting up and optimizing that may take a lot longer.

Thanks Benjamin. I'm not trying to do AFM imaging per-se, but comparing methods for imaging gold nanoparticles on surfaces. The sample should be quite mechanically stable. I had not considered the other factors that you mentioned such as picking up a particle, or shape considerations of the tip. Yes, this is likely material on the surface that might show up on the AFM, although it's very thin and should be under the particles.

An AFM scan you think could image this entire surface (about 1500um^2) in a few minutes, but still give decent resolution at the particle scale? That is very interesting, I thought it would take much longer because the tip has to physically sweep. I guess it can move pretty fast then?

With AFM you will not have to make the surface conductive. The SEM shows the angle of the surface normal with respect to the direction to the secondary electron detector, multiplied by the secondary electron generation rate. The AFM's response is the topography an d the peak height, in addition to elasticity, damping (with the appropriate measurement techniques and for suitable samples). The SEM shows details of the support-particle region as far as it is observable from the side. The AFM is blind there. With the AFM you can manipulate the particles, move them to the side or even remove them. The potential resolution is the same.

Very helpful answers guys, thanks your time

High Adam,

Yes it is possible to image a 100 µmx100 µm or even bigger sized images with closed loop nanopositionning stages. But the bigger the size the longer the time to complete the scan of the area (it is a line by line reconstruction process like old bulky TV) the more thermal drift effect you will get on the slow scan axis (perpendicular to the scanned lines). This might not be of importance in your case as a tip shape "convolution " has to be expected. Diameter of particules may only be assed by a vertical measurement between the substrat and the top of a particle. Regarding this many AMF tip shapes are available to lower the tip shape convolution problem.

As earlier mentionned in this topic, particles adhesion is a problem are sort of. I imaged nanosized gold particles (3-5 nm) on glass surface (non oxyde glass) using tapping mode. Those ones were reported as well bonded to the substrate due to a chemical interaction with the glass. Unfortunately I ended up with a lot of those particles sticking to the probe tip. When lucky this bad is a good and improves you latteral resolution (tip radius of 3 to 5 nm against usually 7 to 10 nm) nonetheless the tip is highly unstable. Yours look bigger may be the adhesion will be high enough to avoid this problem.

Resolution of AFMs is, on the paper, far better than SEM ones (

Why not?

Here are InSb QDs on InAs: 2x2 micron and 1x1 micron (3-D)

The second

Hi Adam,

I agree with previous comments about particles adhesion to the substrate. But once that is solved, it is easy to get nice nanoparticle images using AFM. Here I show 23nm silica nanoparticles imaged in Dynamic mode with a Cervantes AFM from Nanotec Electronica (in both 2D and 3D representations). Acquisition time was a few minutes. If you are intereseted, contact Nanotec (www.nanotec.es) for test sample analysis.

And the 3D one:

Thanks everyone for contributing to this thread. Pablo and Yuri, thanks for these pictures. You said the acquisition time was a few minutes. Am I correct in presuming that the acquisition time scales quadratically with image area? In other words, if you doubled your sample area, would you need to wait 4 times longer for the scan? I only ask because my original image was 10 x 15um roughly, and I wonder how long such an image would require in the AFM.

Thanks for these images. They actually are very similar to what I get from the SEM in terms of image contrast and so forth.

Adam,

Increasing the sample area has different implications for the acquisition time (there are different factors, such as number of pixels, scanning frequency, feedback gains, measuring mode...).

When scanning larger areas, typically it is needed also to decrease the scanning frequency: for example, lets say a 1x1 um image acquired at 1 line/s. The tip will be moving at 1um/s. Keeping the same scanning frequency but increasing the size to 10x10 um, it will move at 10um/s.

In my experience the relation between size and scanning frequency is not linear: increasing the size in a factor of 10 does not imply to decrease the scanning frequency that factor beacuse of the other factor involved.

But for example, if you want to keep a good resolution in large areas when zooming in, you have to increase the number of pixels of your image.

Lets say an image of 10x10 um. In case of a 1024x1024 pixels image, each pixel will have a size of ~ 10x10 nm. If you have 10nm diameter particles you will not resolve them properly. You would need at least 2048x2048 pixels image.Compared to a typical 512x512 pixels image, it will take 4 times longer considering just the number of pixels.

In the end, acquisition time increases, but there is not a clear relation between sizes. Playing with the different parameters (feedback gains, operation mode, scanning only trace...) you would get good images of larger areas also in minutes.

Here you have an example of the whole acquisition process of a 5x5 um graphene nanoribbons image. Image acquisition time after tuning the system is ~ 2-3 minutes:

http://www.youtube.com/watch?v=YgtQ6CcqTZk

Yes, you can reach for sure this resolution, but pay attention to the tip broadening effect. If the particles are spherical you can measure the particles thickness to determine their size (the measurement along the Z-axis is not affected by the tip). You can find in literature different papers or book chapter that describe the tip artifacts and that indicate methods to avoid them.