It’s a problem that few of us will likely ever face: once you’ve built your first homemade integrated circuit, what do you do next? If you’re [Sam Zeloof], the answer is clear: build better integrated circuits.
At least that’s [Sam]’s plan, which his new reactive-ion etching setup aims to make possible. While his Z1 dual differential amplifier chip was a huge success, the photolithography process he used to create the chip had its limitations. The chemical etching process he used is a bit fussy, and prone to undercutting of the mask if the etchant seeps underneath it. As its name implies, RIE uses a plasma of highly reactive ions to do the etching instead, resulting in finer details and opening the door to using more advanced materials.
[Sam]’s RIE rig looks like a plumber’s stainless steel nightmare, in the middle of which sits a vacuum chamber for the wafer to be etched. After evacuating the air, a small amount of fluorinated gas — either carbon tetrafluoride or the always entertaining sulfur hexafluoride — is added to the chamber. A high-voltage feedthrough provides the RF energy needed to create a plasma, which knocks fluorine ions out of the process gas. The negatively charged and extremely reactive fluorine ions are attracted to the wafer, where they attack and etch away the surfaces that aren’t protected by a photoresist layer.
It all sounds simple enough, but the video below reveals the complexity. There are a lot of details, like correctly measuring vacuum, avoiding electrocution, keeping the vacuum pump oil from exploding, and dealing with toxic waste products. Hats off to [Sam’s dad] for pitching in to safely pipe the exhaust gases through the garage door. This ties with [Huygens Optics]’s latest endeavor for the “coolest things to do with fluorine” award.
I never expected the words “Garage” and “Semiconductor fab” to be used in the same sentence…
Looking at that setup, “garage” only describes the literal location of an otherwise well equipped lab. This isn’t cardboard and plywood work.
Very impressed, good work!
If [Sam] shields his garage, he’s a billionaire after the next Carrington-level event.
Am I the only one watching the video who expected to hear the phrase “Excited bromide in an argon matrix”?
Big money college degrees, long student loan payments, scholarships, job security once you get there.
The average Attrition Rate in upper level modern big-name tech sectors is around 15 days to 36 months . There is not a lot of vertical movement within large modern organizations. ;-)
At the big chip company I work at, I routinely am around people in the company for 1, 2, 3, or more decades.
I rather resent this being called a “garage fab”. It *is* an amateur lab, but it’s much larger than a garage unless you are in the auto repair business or a car collector. And it cost more money than most people gross in several years.
I have a 1500 sq ft shop with machine tools and a suite of T&M kit that cost around $500k in the mid-90’s. It took a lifetime for me to afford that and I was paid very well. I didn’t have millionaire parents.
The artificial barriers are significant when building 1 of anything. It is an issue that goes far beyond fiscal limitations, often manifesting as a resource availability and/or a logistical challenge.
i.e. many business people don’t want to bother dealing with small factory orders or lab lots… for example, you probably run into this issue when trying to find domestic material supplies.
North America has mostly adopted a Cargo-cult mentality for STEM.
And tooling up for unspecified project scopes is often just a waste of resources. ;-)
Reminds me of nothing else as much as a university EE or physics lab. Stand back, take another pic, show me the ‘garage’ part, please?
I have a machine in my garage that was $1M new in the mid 2000s that I picked up for less than $5k all said and done. Another that was likely $100k or more that I got for $300. No millionaire parents here. Just thrifty and a bargain hunter, and in it for the long-game.
Very impressive. I wonder how much all of that cost, it was a lot I am sure (commercial fabs cost 10s of billions of $). His feature size seems small but is also still quite large compared to commercial fabs. I did wonder about the legality of venting toxic gases out the garage door although maybe he is under reportable minimums.
no true lab is complete without a *lot* of tin foil.
I wonder how long it will be before someone is able to produce a working fully functional CPU of some sort in an amateur/hobby fab like this?
Sorry, but where exactly is the fluorine exhaust going? Through a door does not make it “safe,” this is an extremely toxic gas. Semiconductor manufacturing requires the use of lots of *very* nasty chemicals, and half of a real fab are built to handle them safely. I hope he’s not skimping on that part.
This looks pretty illegal to me.
In the U.S. release of chemicals depends on Federal, State, and local laws. Usually for airborne release the law is something like “if you can smell it at your property edge, then you’re trespassing on the rights of your neighbors”. Fluorine chemistry is commonly available at grocery stores here (wink rust remover), so it really isn’t that uncommon. Personally for the exhaust I’d want to try some absorber or deactivation, if it was really bad. But it very well might simply react with the air and ozone and degrade with sunlight.
It probably isn’t even close to worrying any legal limit, as the scale is soo very tiny, and the exhaust quite reactive, so it should be in a more stable ‘safe’ state rather quickly and/or disperse into irrelevantly low concentrations rapidly.
That said Its always nice to see some effort put in to making safe the exhaust before it leaves your setup, rather than trusting the world will do it for you. Sensible to make sure you are safe, but rather egotistical to think that is all you should worry about. Is pretty much the default human state though – all those fume hoods you used in school chemistry for instance are usually doing exactly the same thing, just taking the nasty stuff away and dump it ‘elsewhere’.
The problem with statements like “x is impossible” or “y is too dangerous to be attempted” is that they are always based on some assumptions. If someone does not fully understand the problem and assumptions then they often apply these statements in the wrong places. For example, some RIE gasses are deadly but others are commonly inhaled for grade-school science experiments (contrast BCl3 to the SF6 that I am using.) Exhaust products need to be considered for each individual gas rather than making blanket assumptions about all processes. In this vid I used CF4 to etch Si and SiO2 which yields HF, COF2, CO2, CO, SiF4, and recombined CF4 in the exhaust. Among the “worst” of these is HF vapor. NIOSH REL exposure limit is 6 ppm (5 mg/m3) for 15 minutes. My typical runs are 3 minutes and commercial RIE tools that are much larger than mine typically have concentrations of 0.15 mg/m3 for CF4 etching at the exhaust which is 30 times lower than NIOSH REL and already below emission regulation concentrations. Many things become easier and safer in a semiconductor fab if you have fewer small tools and do not care about throughput. I placed HF detector at the exhaust and read <0.1ppm. If one were to build larger tool, process larger wafers, or run multiple RIE machines simultaneously then exhaust product abatement would be a concern.
That may somewhat reflect the difference between hardware and software