• Andrei Markin

Homemade Electron Microscope: What Can This Thing Deux

From the last post I left off having finished building the scanner and spending months desperately trying to improve parts to finally get it working.

The main sticking point was that the scans were coming back as just noise, so for all its worth this scanner may as well be an over-hyped random number generator.

This situation is the most difficult to debug , as each improvement seemed to make no difference to the results. Anyway I'll skip to the final chapter:

On a lazy Saturday morning I fired up the STM, and made a fresh tip just by snapping some tungsten wire. I then took a scan to see what I was dealing with:

The tip moved out of the piezo's range before the scan finished which is why the top half is flat. But the bottom half is interesting.

By this point I have already put in many improvements to lower the noise in my system, and you can see that this image is starting to show some kind of ordered pattern, just distorted in some areas.

For me this pointed back to vibration issues, so I decided to increase the vibration isolation between the scanning stage and is base. A piece of bubble wrap is all I had on hand:

This simple fix was the final nail in the coffin, I re-assembled the scanner and was greeted with this:

I was in such a shock that I resorted to taking a photo of the screen

It works!!!! If I could describe the feeling it would look something like this:

What you are seeing there are individual atoms of carbon! the distance between each one is a tiny 0.14 nm.

To make sure that its not periodic noise; I doubled the DAC steps / pixel and took another scan:

And it works!!! the field of view has become wider. If this was just periodic noise on the detector it would not change with the number of DAC steps / pixel so what we are seeing is indeed the topology of the surface.

You might be thinking; why is the pattern so regular? well this pattern is actually very common in nature. When spherical objects try to pack as closely as possible, they do so in offset rows to minimise gaps, which looks identical to what is seen on the STM image:

As a scientist I would like to put some numbers to the scans, so lets have a look at how close was my initial estimate of the piezo actuators movement:

From the experiment with the optical microscope I guesstimated that the tip moves around 200 nm / V. The DAC was setup with a 5 V reference and 2^16 steps which gives 76 uV / step, so combining these two numbers means that the tip should move:

(200 nm / V) * (76 x 10^-6 V / step) = 0.015 nm / step.

Here is a closer look at the first image. In crystallography such patterns are described by a unit cell (the black box). Its dimensions (a) are known to be around 0.14 nm for HOPG. In my scan this is around 12 steps, so 0.14 nm / 12 steps = 0.012 nm / step.

So that initial guesstimate was actually incredibly accurate.

Now that the scanner is giving images we can work with, lets see if its possible to reduce distortions. Here is a scan I did with a fresh tip:

You can see that the entire left side is badly distorted and the bottom is also not great, but the top right shows a nice close packed structure. The x and y scale shows that these distortions are at the bottom of the DACs range, and parallel lines suggest that the piezo voltage is not increasing linearly as we step through the DAC ladder.

To correct for this I added a constant offset to the DAC so that it is never at the bottom of its range:

The tip moved out of the piezos range at the bottom

This did improve the image so no more long horizontal streaks.

But now its clear that i'm getting some kind of periodic noise distorting the scan (vertical lines near start, middle and end). The cause of this issue took quite a while to diagnose but since I have definitely achieved what I set out to do, this can be the subject of a future post.

Finally I wanted to have something physical to show for all my hard work in getting this to work. So I wrote a short MATLAB script which can take the surface data acquired by the STM an press it into a stereolithography file, which can then be 3d printed:

On the left is one printed on my home FDM 3d printer and on the right is one printed on a high resolution SLA printer (form labs, form 2)

If you would like your own then feel free to 3d print one yourself.

This is not the end for this project, I've still got several upgrades in the pipeline and many ideas of what to do with it, one of which is to incorporate my masters project and perhaps measure the conductance of individual molecules.

204 views0 comments

© Andrei Markin 2019

This site was designed with the
website builder. Create your website today.
Start Now