Saturday, March 26, 2011

DIY Scanning Electron Microscope - Sources, Costs and References

Metal aluminum window screen

Just a few linearity problems ;)

I used an oscilloscope's X and Y amplifiers for these images. It has much better linearity than my own, but not enough differential voltage or offset range.

The sum total of the big-ticket items shown in the video is $1485. This does not include hoses, wiring, raw metal, teflon, screws, a cabinet, etc. It also does not include an oscilloscope, which can be a very simple model (under $100 on eBay) as long as it has a z axis (brightness) input. Your diffusion pump or diffusion pump baffle may also require a water chiller.

Here are a list of information sources that helped me with this project:

Teralab - Homebuilt electron gun and other great projects

Popular Mechanics video on commercial desktop SEM

Hamamatsu - Supplier of PMTs

TV to oscilloscope circuit

CRT oscilloscope clock circuit

Charged particle optics simulation program

"A Simple Scanning Electron Microscope" P.J. Spreadbury -- Advances in Imaging and Electron Physics Vol 133 Chapter 2.5 (no link).

ISI SEM refurb at home

Great technical info on cathodes and wehnelt cup spacing. Most of the article concerns LaB6 cathodes, but there is a short paragraph on tungsten cathodes.

Numerous websites that gave background and operational information about SEMs.
Lots of web searches

Nearly all raw materials for this project were purchased from McMaster-Carr. All power supplies were purchased on eBay, or I already had them, in which case they came from a surplus store or flea market. Nearly all of the electronic components came from Jameco.


  1. Heh. I must be one of the first people to see this video because I went and picked up that spellman you mentioned for $70.

    Nice. I have some neon transformers that I modified to be about 12kV DC with some microwave oven diodes for another project, but those are unregulated so probably won't work well in this role. (however, they are dirt cheap).

    If you want some cheap phosphor screens, the bell jar ( sells sealed plastic phosphor screens for $18 each that are sensitive to electrons and x-rays.

    Something I wondered, I have a jar of glow in the dark paint, do you think just painting it on a piece of glass would work as a cheap phosphor screen? It seems to be very sensitive to UV light (my UV laser can "draw" on it and leave a glowing trail.) so perhaps it can be excited by the electrons directly or the bremsstrahlung radiation of them hitting the screen, which is not all that far from UV in your beam voltage range.

  2. John, the phosphors used in Everhart-Thornley detectors in SEMs are very "fast". The light pulse emitted after being excited by an electron lasts about 100ns. If the phosphor were slow, the output light pulse would be long, this would limit the microscope's ability to scan the sample in a reasonable amount of time.

    There is a difference between phosphorescence and fluorescence. One term refers to a material property in which the material emits light upon excitation, and the other refers to a material's ability to retain energy from excitation and release light slowly over time. I forget which is which -- check out wikipedia.

    You could add voltage regulation to your neon sign xformer. All linear high-voltage supplies probably make use of a very similar xformer. I recommended against flyback xformers, because they are so spikey and difficult to deal with.

  3. Facinating! The company I work for ( just scrapped an ancient Hitachi TEM that could also be used as an SEM. We actually had to take photographic images of the phosophor screen when using it as a TEM, although I think there was also the possibility of direct electron exposure of film plates. That meant we also had a dark room, chemicals, enlargers, etc. that were also scrapped. If I had room for it, I would have snuck it all into my garage when my wife wasn't looking... it was finally given away to someone who agreed to disassemble it and carry it off.

    I doubt you need worry too much about it at the voltage and currents you are using, but an SEM also produces x-rays. You should check with your state health department to see if it needs to be licensed. Our TEM/SEM was licensed. In fact, with appropriate liquid-nitrogen cooled "magic crystal" instrumentation you can add energy dispersive analysis by x-ray (EDAX) to your imaging capabilities. I thought that was way cool when I first saw it thirty years ago on a commercial SEM: an image of how the elements were physically distributed on the surface of an alloy sample! It's a quick way to identify the chemical composition of unknown samples of, say, moly or tantalum or whatever.

    I like your vacuum system. We use a lot of oil diff pumps in my lab; I have two Varian 6" pumps on my tandem accelerator, along with a couple of turbo pumps at the ion target end station, where a good vacuum really counts. I also have a small e-beam hearth in a glass Bell jar, only slightly larger than yours, that is pumped with a diff pump, trapped with a small LN2 cold trap. You only need about 10^-5 torr for e-beam heating, but the diff pump gets there really quick. BTW, an oil diffusion pump does have moving parts: the supersonic molecular jets of oil vapor that transfer their momentum to move air through the pump.

    Keep up the good work and have fun!


  4. Are you actually using separate high voltage power supplies for all the areas that need high voltage? I would think a lot of those voltages could be produced with voltage dividers. The power for the deflection plates might need something more stable.
    Looking forward to future videos.

  5. Anon, my first plan was to use potentiometers for the focus and maybe Faraday cage voltage. The problem is finding high-value, high-voltage potentiometers. With a 10KV supply, a 10Mohm pot would draw 1mA, and dissipate 10 watts. This would not work, so I really need a 100Mohm pot that can handle 10KV -- very difficult to find. I bought a flyback transformer for a color TV with a focus control, and it indeed contains a high-value pot, but it is purpose-built, and very difficult to remove since it is potted in epoxy and contains other fixed-value integrated resistors. I found it easier and cheaper to just buy the supplies for each voltage that I needed.

  6. I'm not a connoisseur on the matter, but I have to say that this is an absolutely stunning project!

  7. I have found all "Advances in Imaging and Electron Physics Vol 133" book on the below link:,%202004%29.pdf

    Plenty of information on Sir Charles Oatley and the scanning electron microscope.

  8. Hi Ben,
    your work is really inspiring. Designing my HV-PSUs from scratch right now...
    Just a quick question: Are you russian? (suggestion based on your name) no offence ^^

  9. artemonster, thanks. My name comes from Ukraine where my father's grandparents lived.

  10. One question: how does the voltage ripple on accelerating PSU will affect the picture quality? Is stability for accelerator so crucial, as for example for electrostatic lens? As I've said, I'm designing a set of HV-PSUs right now, but in Internet there is absolutely no good specifications for PSUs for SEM. Can you advise some good sources, where good technical information could be found? Thanks.

  11. and also another few dumb questions: is there any other advantages of magnetic deflection and magnetic focusing over electrostatic, except for not using HV-PSUs and just current sources? Any disadvantages?(hard-to-make or else?) Thanks.

  12. artemonster, all power supplies in the SEM should be very stable. Ripple in the accelerating voltage will cause image distortions because electrons with higher velocity will be less affected by the focusing and deflection elements than slower electrons. If the accel voltage has a lot of ripple, the scan pattern will not be a clean square/rectangle.

    To my surprise, the tungsten filament must also have a very stable DC power supply. I originally planned to use AC, and there was significant 60Hz noise in the beam. I think the filament might have been physically moving due to the AC, or the very slight 3V bias caused by the filament voltage could have been casing a problem.

    Magnetic focusing is usually preferable over electrostatic because the magnetic fields can be made more intense, and provide more beam control. In addition, coils of wire take high currents at low voltages, which is sometimes easier to provide than high voltages at low currents required for electrostatic. Unfortunately, magnetic focusing requires precision iron pole pieces to be machined and coils to be carefully wound. Electrostatic control only requires a metal plate or cylinder, and making changes/modifications is easier.

  13. Gabriel Mendez, Arpaseternas@gmail.comOctober 18, 2012 at 7:29 PM

    Why not to use a needle as very sharp point electron field emitter?.
    I believe that you can obtain a dozen of uAmp.

  14. Gabriel, making a field emission source would be a fun project in itself. They require better vacuum, and the dimensional tolerances are very tight. Tungsten filaments are easier to use and more robust, which is why I chose to use them.

  15. Hello your using electrostatic lens?
    At what voltage do you use at the condenser lens?

  16. Anonymous, yes my SEM design uses all electrostatic beam control. The condenser lens is held at the same potential as the electron gun itself (between 1 and 10 KV).

  17. Congratulations.

  18. Ben, great project, well-described in the videos. Instead of doing my homework in high school I built oscilloscopes with home-brew power supplies-- buying the power, regulated, on Ebay lets you get on with the real fun.
    -- Is a photomultiplier tube (my college physics summer job) really necessary to detect the emitted electrons? This just shows my age, but I'd have thought there would be a more direct detector with the required sensitivity.
    -- Altogether very cool. Thanks.

    1. Phil, thanks for the message. The PMT+scintillator approach seems convoluted and I searched for a long time to determine why this is the best approach. The main benefit is that the sensor surface (scintillator) can be at a very high potential (+10KV) for high sensitivity to electrons while the PMT anode can be held at ground for easy connection to the rest of the visualization circuit. This allows effective DC coupling between the 10KV potential and the SEM output signal. This could also be accomplished with digital electronics, isolated supplies, fiberoptic communications, etc, but the PMT+scintillator is relatively cheap, easy, and offers really good performance. Mass spectrometers use direct electron multipliers (instead of PMT), but these need to be setup with the anode at a relatively high potential, and need to be AC coupled or sensed with isolated, floating amplifiers.

  19. Ben, Great project. Can you tell me where to buy the electromagnets for these microscopes? I am working on building a TEM and I couldn't find any manufacturers who does these kinds of high power electromagnets.

    1. Sujay, you are correct that finding magnetic "pole pieces" or complete electromagnets for microscopes is very difficult. This is why I built my microscope with entirely electrostatic parts. I used no electromagnets, and only a couple simple permanent magnets for beam alignment. Electrostatic beam optics do not offer as much control as electromagnetic, but is much simpler to build.

  20. Your videos are quite fascinating! I have watched all of your SEM videos and I am building a similar device but for high energy electron characterization (10x higher voltage than yours). I hope you wouldn't mind if I asked you a couple short questions?

    I have been doing a literature review looking for some good approximations for electron beam divergence to figure out where to place my lens along the column. I don't have any electron optics software available to me to run a calculation. Also, I would like to know if there is a specific reason you chose copper to form your channel, being that it is conducting and opens the possibility of arcing within?

    I'm grateful for any help!

  21. Hi there, any plans of making TEM? I'm very keen to make one! Thank you