[Ben Krasnow] is quite possibly the only hacker with a Scanning Electron Microscope (SEM) collection. He’s acquired a JEOL JSM-T200, which was hot stuff back in the early 1980’s. [Ben] got a great deal, too. He only had to pay shipping from Sweden to his garage. The SEM was actually dropped during shipment, but thankfully the only damage was a loose CRT neck plug. The JSM-T200 joins [Ben’s] homemade SEM, his DIY CT scanner, the perfect cookie machine, and a host of other projects in his lab.
The JSM-T200 is old tech; the primary way to store an image from this machine is through a screen-mounted Polaroid camera, much like an old oscilloscope. However, it still has a lot in common with current SEMs. In live video modes, an SEM can only collect one or two reflected electrons off a given section of a target. This creates a low contrast ghostly image we’ve come to associate with SEMs.
Attempting to fire more electrons at the target will de-focus the beam due to the electrons repelling each other. Trying to fire the electrons from higher voltages will just embed them into the target. Even SEMs with newer technology have to contend with these issues. Luckily, there is a way around them.
When “writing to photo”, the microscope switches to a slow scan mode, where the image is scanned over a period of a minute. This slower scan gives the microscope extra time to fire and collect more electrons – leading to a much better image. Using this mode, [Ben] discovered his microscope was capable of producing high-resolution digital images. It just needed a digital acquisition subsystem grafted on.
Click past the break to see how [Ben] modernized his microscope!
Using the schematic, [Ben] connected his Tektronix MDO3000 series oscilloscope to the video signal of his SEM. Tek was kind enough to give this model to [Ben], [Dave Jones], and several other prominent hackers. We’d love to try one out too, but we’re pretty sure we’ve been permanently placed on their naughty list.
By triggering on the SEM’s vertical refresh, [Ben] was able to capture an entire image into the scope’s memory. He exported the data to a USB stick and loaded it into GNU Octave, He wrote a simple script to search for the horizontal refresh pulse and build up a raster image. The results are stunning, to say the least. [Ben’s] was able to capture the classic fly eye without first plating the fly with metal, as is normally required for SEMs.
We can’t wait to see what [Ben] comes up with next!
Whenever I see a new article on Ben’s blog, I know it will be cool (“homebrew LCD”, “DIY SEM” cool). Haven’t been disappointed yet :)
I work with Electron Microscopes on a daily basis and i pity those who don’t.
EMs are so fascinating machines. Sitting in a climated, completly dark room, lit only with the faint green light of electrons striking a phosphor (yes, i know its aluminium doped zinc sulfide) screen. You sit in front of the machine, holding your breath, moving carefully to avoid bumping into the bench, slightly adjusting the stigmator coils to correct the abberation of the objective lense and pray for the next 4 seconds of film exposure that the specimen does not drift away. However, 99 % of the pictures are crap. But at the end of the day, there is the one picture… taken with atomic resolution… and you get this feeling… The feeling of “BOOYA!”
Seriously.. search for microscope facilities near you or go directly to your local university to the physics, engineers or solid state chemistry department and kindly ask for a demonstration of these machines. Most scientist will happily show you their microscopes in action. Maybe they have an old or dying machine and are willing to give it away for small money (0,25$/lb).
Kindly asking gets you places. ;)
I work at a place that has a SEM. While it’s not my job to work *with* it, I take care of the people who do; fixing their computers, and on several occasions being given boards from the SEM itself to repair. So one day I did what you suggest, and asked for a demo.
After a quick and successful search for a fly, I got it.
While the fly was being readied and the chamber evacuated, I was given a crash course in basic theory. [Adam] touched on much of it. The beam adds electrons to the target, and if it accumulates too much of a static charge, it then repels the beam. So the target is mounted using low-outgassing conductive carbon tape, allowing it to be grounded. And it’s best if the target is inherently metal, or metal-coated. But a fly is at least a little conductive, and so lower-quality realtime images are still possible at low beam energy.
Then I got sat down, shown the controls, and allowed to run it. It was an old-timey SEM with the zinc sulfide screen. They’ve since upgraded to a new state-of-the-art model that’s completely computerized, but I’m glad I got to use the old one instead, it made you feel more connected with the machine. What a wonderful experience.
I played with it for a half hour. There was one big surprise. At one point while experimenting with how scan size/rate/power affected image quality, I had the beam focused quite tightly, and noticed the image progressively changing even though I was no longer adjusting the controls. Zooming out, I found I’d created a tiny depression in the fly’s eye. After asking to make sure I wasn’t contaminating the chamber with eye particles (ha!), I had a little fun with my discovery. I still have the fly, which now bears my name, written microscopically on its eye.
Yes, if you ever get the chance, take [Tobias]’s suggestion to ask kindly for a demo.
Ah yes, the infamous ‘stigmator’ controls.. a royal pain in the ass when trying to focus at magnifications in excess of 50,000x.
I LOVE me some SEM time. It’s like sitting down at a portal where you get to look at alien worlds for a couple of hours. SO AWESOME.
How does he even have time to do this after his Google hire? Does he sleep?
What? What have i missed?
https://plus.google.com/+BenKrasnow/posts
Works at Google [x]
Using the scope is a neat trick, however sadly I’m unlikely to ever own such a scope. I wonder how good an image he could get with a CHDK hacked camera set for multi-second exposure.
I’d say not as good as the technique he described in great detail in the video.
you dont need a fancy scope, you need a digital acquisition system. He wasnt even sampling with an outrageous frequency (was it something like 2MHz?).
It actually works pretty well. This is from the old SEM we found on craigslist a couple years ago:
https://www.flickr.com/photos/67292116@N00/7201775882/in/set-72157629736821312
We have been trying to figure out a much cheaper and more straightforward way to do what Ben is doing. That scope he has is about $13k.
If anyone in or around Chicago is interested in SEMs – Pumping Station: One has a fully functional scope.
http://pumpingstationone.org/?s=SEM
This is sooo cool! I want one!
At the slowest scan settings it looks like a soundcard would provide a high enough sample rate to capture the image. Low frequency roll-off because of the input DC blocking would be an issue though. Maybe a modded soundcard and something to attenuate the input signal would eliminate the expensive scope?
Ben talked about that in the comments on the video. He initially wanted to use a cheap USB sound card, but the chipset in it used internal DC input decoupling. So he sought some other method and used the MDO.
If you found a card with no chipset-internal decoupling, just external, yes you could modify it by removing the offending caps and it would probably work just fine.
A bit hackish, but a low frequency signal can be chopped or amplitude modulated with a high frequency signal, to get it into a DC-coupled soundcard. Then filter in software.
Even an on-off chopper would work for half-the amplitude.
Yeah, I think HaD might be on Tektronix’s “naughty list” as well as the other 0-scope company as well… Kudos to you!
It’s really frustrating seeing this type of ludicrously-expensive kit being used on a DIY blog like Hackaday. A $13K+ scope is beyond the reach of virtually everyone.
And?
Everything Ben does is worth covering on HAD, at least briefly. Some are more affordable than others, but it is all proof that you can do these things at home. And face it, it is cool as hell!
You may not be able to afford the scope, but you could build an acquisition system for this specific task for well under $100.
Or as mentioned, use a soundcard. I don’t think you’d neccesarily need to DC couple it as it should be possible to dc-clamp it in software afterwards.
A generic soundcard won’t work, it just won’t pass the DC signal through. There are workarounds though.
Yes, its as frustrating as passing expensive cars, or walking past a nice house.
wait WHAT? are you autistic, or just trolling?
Out of reach – for now. Sooner or later someone’s going to RE a SEM until it is completely salted and tell everyone what it is that makes it tick (guessing focused e-beam generation) so we can start hacking our own. Then they’ll start appearing in prominent hackerspaces.
The basic theory behind an SEM isn’t that complicated (it’s taught at A-level here in the UK, ie high school level).
As you guessed it’s basically a beam of electrons focused by magnets onto a target, and the reflected electrons are picked up with some kind of detector. In older ones such as this one the detector is a phosphorescent screen.
The difficulties would be that the sample and beam need to be kept in a vaccum, and the beam needs to be very well focused.
Time for you to resume hacking that CRT?
This guy is an American treasure.
Ben isn’t the only one with a SEM in the garage – Jeri has one too.
wow amazing images.
You can do this with a simple camera. I use a Hitachi S-800 SEM with a Canon DSLR and a 100mm macro lens on manual focus (see photos at [1]). Right now the camera is tripod mounted, but I am remaking the original 4×5 polaroid camera back and bellows to allow direct mounting of the DLSR onto the face of the display panel. I have a bulb shutter release for the camera. You get the image you want and then hit the Photo button on the SEM and open the camera shutter. When the SEM sounds the tone to indicate that it is done with the photo, you release the shutter. The remote shutter control has a locking position so I do not need to hold it. The image is automatically downloaded to a computer and put up on the display. It is all very simple and easy.
Thanks,
Bryan
[1] https://plus.google.com/photos/114736545683009543907/albums/6055179846602206097?authkey=COvX0M-U4YOxpgE
I acquired some time ago JSM T-20 with set of service parts for 150$. It looks very similar to T-200, but has second CRT. I’m trying to find service manual but so far failed.