So I recently championed the acquisition of a nice TIG welder at the Heatsync Labs hackerspace (special thanks to Karl at K-Zell Metals for getting us such a good deal). I want us to always be able to make anything, even space ships, so I want us to have the tools we need, be it laser cutters or TIG welders. However, my real reason for wanting a good welder was to weld vacuum chambers, fittings, tee’s, etc.
I’ve since discovered that I’m not very good at welding. But I’m trying. Which illustrates exactly my passion for this project! The SEM project truly encompasses every conceivable form of fabrication that I can think of. Digital electronics, power electronics, high voltage, RF, analog front ends, welding, machining, vacuum physics, mechanical assembly, materials science, particle physics, electromagnetic software design, human interface systems… the list goes on and on and on!!
Literally everyone who walks into the hackerspace has something, or more realistically, multiple things to offer this project. I have learned so many skills by passionately chasing this dream around for the past two years and for that I am so thankful! I am thankful that a creative community such as Heatsync Labs, and the entire maker movement in general, exists. This is something that didn’t exist when I was little, and to be a part of it is awe inspiring.
I kicked around a few ideas about how to build a free standing “developmental” version of the gun, with an improved rigid support structure. Also shows is an all-metal version which should prove to be better as a final version, though it sadly hides the entire discharge tube from sight and therefore is not a good candidate for a “development” item. Call it future idea #7633…
HDPE is not the best material for a high vacuum, but it does OK under low vacuum…
Time to make things better. The first prototype cold cathode electron gun used a piece of aluminum foil as a beam aperture. Talk about a not at all precise piece of optics. The new Mark I gun uses a lathe turned rubber stopper to support a piece of lathe turned tungsten and a small peiece of #32 AWG wire to bias a lathe turned aluminum anode with a 0.46mm aperture.
Interestingly enough the electron spot is much larger even though the length of the exit aperture is longer (more tunnel, less hole in foil) but it IS larger in diameter. Not to worry because a larger divergence (high optical numerical aperture or N.A.) actually means a smaller spot size once we start using the lenses to focus things down. But it makes me wonder if there is some scattering off of the inside walls of the aperture bore. Hmm…
But there are still problems. The rubber stopper is very difficult to center and likes to wobble. Its impossible to keep the electron gun on center and the spot wanders because of this. More than once the glass tube supporting the gas discharge fell off inside the beamline and had to be fished out after damaging the phosphor coated detector. Also, the aperture is jam fit to the glass and has caused the glass to crack more than once. Its a bad idea, the whole assembly really needs the be rigidly attached. Mark II is on its way already… But its still progress, and its still really pretty to see!
Oh, and please ignore my crappy attempt at a beam current detector. It really didn’t work at all. Which makes sense considering how low the beam current is.
An Awful Horrible Deliciously Hacked Up Electron Lens
Its not pretty. (Is… is she using a… harbor freight clamp to hold the lens at the focal point? yikes…) I’m sorry, because, it is in fact indeed awful and horrible and hideous. Its arguably uglier than my current divorce… (hilarious! … *crickets* …*clears throat*) But you know what it is?
Its a gap-solenoid electron lens! Its a baby step, and I might even go so far as to say that it is even more than just a baby step because it is an electron spot, and an electron spot is exactly what an SEM needs to be… well… an SEM! But its also a low-vacuum electron spot and totally affordable and hackable electron spot. Its.. its progress!
The lens consists of a piece of lathe turned 1018 steel. Its a spool shaped central piece which is cut in half to allow for a small 1mm thick brass washer to be inserted. The 500 turns of #20 AWG magnet wire is wrapped around either side of the gap washer and a sleeve of lathe turned 1018 steel is slid over and clamped onto the spool shape. The end result is a bit like a hollow donut with 0.070” thick walls and a rectangular cross section instead of the familiar circular one we all know and love and dip in coffee. Except its steel.
The resulting focal length is about 6 inches with 800mA of lens current at the 10kV acceleration voltage. The actual beam energy has yet to be measured and is surely less than this due to the voltage drop across the plasma used to create the beam. This is probably somewhat excessive and the number of turns should be doubled to reduce lens heating. But it works for now so I don’t really want to touch it.
Please bare with me, this narration is coming from the past. I try to always take pics and document everything but life happens and the blarg doesn’t get updated and then I have to do this :)
Using what I saw on Mr Styner’s amazing and truly humbling website, I constructed a simple experimental electron gun using a copper wire in a capillary tube, a rubber stopper, and a turned stainless steel needle. Its really not even worth describing what I did because it looked horrible and didn’t even work. It did make a nice discharge tube tough. Mmmm delicious nitrogen plasma… Several things were wrong, the polarity of the chamber vs the gun caused the plasma to extend all the way down the beamline. Also, it was an unfocused electron spray, not really a directional spot.
SEM filaments are biased at a large negative potential with the anode and specimen placed at ground potential. 1-20kV is a reasonable range for the bias voltage. However, the tungsten hairpin filament requires a large current, several amps. So how do you bias a current source at such a high potential?
Step 1 Find a huge transformer
A good rule of thumb is that the power of a transformer core is the square of the cross sectional area in cm^2. Since I planned on cutting off the center leg, and the side arms measure 2 cm x 5 cm = 10 cm^2, the power core leg is approx 10*10 = 100 VA.
The place of choice for this in the Phoenix area is Apache Reclamation. This place is seriously awesome. Much like a dusty version of the borg cube one can easily spend HOURS inside the poorly lit and dangerously cluttered back room. The transformer shown here was about $15 and I picked it because the laminations are bolted and not welded/riveted.
Knowing how much air is in a vacuum chamber is really important. You can keep an eye on possible leaks, outgassing, but you also need to know if you are going to break something. Tungsten hairpin electron emitters will burn up at pressures greater than 10^-2 torr, a diffusion pump cannot be turned on until 10^-3 torr or fire will happen, and a sputtered ion pump works best at 10^-6 torr or better.
Our chamber came with a DV-6M thermocouple gauge. The gauge is made up of thermocouple wire which is smaller than a hair. A small current heats the wire, and any air will carry the heat away. At 10^-3 torr the filament will reach a stable 300°C operating point. It is meant to be AC excited and then use an analog meter to read the DC offset from the thermocouples. The meter will be unaffected by high frequency AC (the needle just can’t move that fast) but for a digital system this is problematic.
An alternative is to use an interruptable DC heater current and then measure the DC offset during the short (1ms) heating pause. I came up with a schematic which does this. Not shown is an instrumentation amplifier to boost the thermocouple voltage to a level which can be read by an arduino ADC and reported back to a PC.
It was decided that we needed a major overhaul to the chamber anyway, so instead of fix the existing valves we decided to drill out all of the holes and TIG weld real vacuum flanges onto it. We found a welder, Claude, at Mesa Community College who was willing to help us out and in fact volunteer his time as long as we paid for consumables (shield gas). WOW! Thank you SO MUCH Claude!!
But before we can weld, we needed flanges. $240 was raised by Heatsync members (you rock!) to go towards the $318 of flanges that were required to totally, once and for all, pimp out the chamber. We really didn’t want to waste Claude’s time by having him weld pieces every so often.
Total cost breakdown was
ANCORP Part Qty Price QF25 AL LONG WELD FLANGE 6 $84.00 QF40 AL LONG WELD FLANGE 1 $15.50 QF50 AL LONG WELD FLANGE 1 $24.00 QF25 AL WING-NUT CLAMP 6 $45.00 QF40 AL WING-NUT CLAMP 1 $9.00 QF50 AL WING-NUT CLAMP 1 $15.00 QF25 AL CTR RING W/ VITON 6 $39.00 QF40 AL CTR RING W/ VITON 1 $8.50 QF50 AL CTR RING W/ VITON 1 $11.00 QF25 AL END CAP 6 $36.00 QF40 AL END CAP 1 $10.50 QF50 AL END CAP 1 $11.00 QF16 AL LONG WELD FLANGE 1 $11.00 PVC VACUUM HOSE 3 $16.80 $336.30
Duniway Stockroom Part Qty Price KF25 AL Hose Adapter 1 $13.00 KF40 AL Hose Adapter 1 $17.00 $30.00
Step 1 is to create a vacuum. Obviously we need a vacuum chamber to do this.
Heatsync raised $400 in order to purchase this surplus aluminum vacuum chamber and I fronted the extra $400 required to cover the chamber and the dual vain rotary pump. THANK YOU to everyone who pitched in to help!!
A course vacuum was achieved within days, however it was discovered that there was a leak…. When we attempted to fix the leak, the main valve of the chamber sheared off inside the chamber…
Has anyone at Heatsync Labs ever built an SEM before? No. Have any of us worked in medium or high vacuum technology? Not really. Is that going to stop us? hah! Just shut up and hack.
On Feb 17th I gave a talk at the HSL biweekly meeting along with Brandon Guida. Brandon works in the ASU Microscopy lab and they have SEMS, TEMS, and all sorts of other EMs there among other things.
In our talk we tackled the hard question. What the hell IS an SEM and how do we get around to making one.
Question: What is an SEM? Answer: An SEM is a glorified TV. Backwards. Response: .. Oh .. Well what’s in a TV?
So lets go a little deeper. The whole point of an SEM is to get a tiny little electron dot. That’s really the end goal. The brighter and tinier that dot, the better the microscope. Then you just move around the dot while keeping track of where you moved it and it becomes a rasterized image (once you get around to detecting the dot that is). You can’t get a dot without a vacuum so the question breaks down futher.
How do you get a vacuum? You get/build a vacuum chamber and you pull all of the air out of it with a vacuum pump. How much vacuum do you need? As much as you can get. Startup cost on vacuum chambers is expensive unless you can machine large setups and TIG weld them too.
Ok so now we have a theoretical vacuum. How do you make an electron dot? Well first you need an electron beam, which fires out of an electron gun. And an electron gun is just an electron source (like a hot lightbulb filament) with extra bits to make it directional.
How do you make the electrons directional? How do you get the beam down to a dot? Simple. Electron Lenses. Its a deep topic, but electrons are charged particles and they can be influenced by electrostatic charges as well as magnetic fields. The same idea can be used to deflect the dot so that it translates across the specimin to create the raster image.
Uhm, ok now we have a vacuum with an electron beam in it and its being focused down into a dot that’s moving around. Where does the image come from? The image comes from either backscatter or secondary electrons which spit back OUT from the dot and need to be detected. For this we use an electron detector. Think of a solar cell, or a reverse biased PN silicon junction.
So to get the image, we “watch the dot” with a single pixel and record the intensity of it in electrons instead of visible light. Pull the data into a computer and plot. Done. Easy, right?
Welcome to the buildblog for the Heatsync Labs SEM. Let’s backtrack a little bit and bring everyone up to speed.
SEM is short for Scanning Electron Microscope. And a Scanning Electron Microscope is capable of some truly amazing imaging. Just google image it.
SEMs start at around US$70,000 and can rocket into the millions of dollars. From tiny tabletop units to massive beasts capable of simultaneous imaging of multiple areas on full sized turbine blades. This kind of cost sort of makes it difficult to use one.
Why are SEMs expensive?
1.) They are research devices. Noone realy needs one to be cheaper. You just write it into the grant and off you go. Even the “cheap” 70k desktop unit is really just a curiosity among the professionals.
2.) High vacuums are hard. Let me rephraise. They’re hard to make maintinance free. Which means you get a maintinance guy to come fix it all the time. And he charges a LOT of money.
3.) Most SEMs use old 70s or 80s analog parts with 90s or modern era digital parts piggybacked. This increases complexity, power consumption, and size. Which means more $$.
4.) The ones that are truly modern can now be marketed as a “feature”. That’s cool, if you have $2 Million bucks to blow.
So why can’t we make our own? Is there really anything from stopping us?
Let’s ponder that in the next post and we can go over some theory as well.