MEMs oscillator sensitivity to helium (helium kills iPhones)
Applied Science on why helium kills iPhones

Follow by Email
I tested the 32KHz MEMs oscillator used in some phones, and discovered that just a 2% helium environment will cause the oscillator to fail. I also do a teardown of the MEMs device, and inspect it under the electron microscope. iFixit blog post: and vid: Original thread about MRI facility and iPhone failure on Reddit: MEMs silicon encapsulation: I added images and raw data here: Applied Science on Patreon:


NurdRage : holy crap that's interesting.

John NZ : The ultimate punishment for teenagers, if they don’t behave put their phone in a bag of helium, and they can’t use their phone for 3 days...😂😂

J-Hawks : Hilarious joke, at the next apple keynote, flood the stage with helium ... epic lulz

Brian Park : There is no mystery for the slow recovery! I would assume only millitorr of He in the device will kill it. So with 2% He outside (15 torr), the difference driving the diffusion is 15 - .001 ~ 15 torr. Now we have the device filled to more then .001 torr (and it has failed). Now we put it in essentially 0 torr atmosphere, and the diffusion driving the He out is only 0.001 torr! No wonder recovery is so slow! (Perhaps it is not so extreme, as I assumed only 1 millitorr failure pressure). I am familiar with this process as I used it to refill old HeNe laser tubes. Glass (especially pure quartz laser windows) is a "sieve" for helium. Operating He pressure for the tubes is about 1 torr (Ne 0.1 torr). I used 0.1 atmosphere He partial pressure outside the tube to do the refill; it takes several weeks. (I use low pressure to slow the fill, and avoid arc-over outside the tube when testing). If you overshoot, you must wait years for He pressure inside to reduce! Most quartz crystals will work in 1 atmosphere. This oscillator technology must be very marginal to fail at such low pressures! This is miniaturization gone too far! I'll take the big can!

Alexander Guryanov : Wow! Very interesting. Thank you for this information and your investigations.

keeperofthegood : Look up the paper: Diffusion in Silicon Written by: Scotten W. Jones. In that it is stated that He has a constant of diffusion that is a lot higher than Hydrogen. From that paper, Helium diffusivity (constant of diffusion) is 0.11 cm^2/s, Hydrogen's constant is 9.4x10^-3 cm^2/s but there are other compounds listed and some (like S 0.92 cm^2/s) are much higher than He, it would be interesting to use a sealed chamber and hot plate and liberate some S into the environment of these chips and see what the failure process would be and if it is recoverable.

webnothing : Every one of your videos is gold. They are extremely fascinating. Thank you.

Eric Thornton : "At least its kpa" lol

Chris Sloan : An interesting idea to try would be to see if you can restart the oscillator after exposing it to helium, but before it has stopped on its own. The way your test works, you have it oscillating and then you introduce helium. So, it is already oscillating as the helium concentration increases. However, it is possible that it takes more helium to "kill" it than it does to prevent it from starting. Normal oscillators are like that with cold. If take a running computer and drop the ambient temperature at some point it will stop running (0 C, -20 C, -40 C, whatever, it depends on the computer and the components they used to make it). But before that point, if you were to turn it off and back on it would fail to start. Maybe it's the same with the helium. That would explain why it takes days for it to recover. If you were to turn the oscillator on and off as it sits in helium, I bet there's a decent chance that it would fail to start sooner, but it would also recover sooner after that point because less helium has been absorbed.

micah lucas : the geometry of that mems is blowing my mind. Is there anyway to view the electron microscope images in higher resolution?

Tom : Is this why my phone stopped working after I left it in my tokamak?

xMotoMitchx : Interesting! I work in the semiconductor industry, and we use helium to leak check systems under a vacuum. Good to know!

klasop : But I don't understand what is the mechanism how the helium makes the MEMs device to fail but not the hidrogen. :(

evilcanofdrpepper : Every time I see an Applied Science video in my feed I get excited because there is a 90% chance that I'm about to see something super awesome!

Michael Kaliski : A stunning bit of work and hearty congratulations. Helium does leak through any and every containment vessel, so this all makes perfect sense. The timing difference between a device becoming disabled and reactivated by Helium diffussion is almost certainly due to differential vapour pressure. Initially there are billions of Helium atoms trying to diffuse into the MEMS device. However it only takes a few million molecules to stop the device working. When the external Helium atmosphere is removed, there are only a few million atoms to diffuse out, so much less impetus propelling the Helium molecules back out from the MEMs into the atmosphere. The Silicon containment provides sufficient protection from a vacuum to prevent that from speeding up the recovery process on its own. Perhaps the application of moderate heating combined with a vacuum might speed up the process slightly.

24680kong : It would be interesting to see if you could open up a small hole in the MEMs chip so that hydrogen didn't have to just rely on permeation to get in. Then you could see if it still affects the oscillation. You could then determine if it's just the slow permeation that makes hydrogen unable to kill the iphone or if it also can't get into the oscillator to cause the crash.

kerajit : Applied Science == like

The Lamb : I love the in depth details that you give. On many occasions they go over my head but I managed to stay with you on this one. Really fascinating production method. Cheers.

Jared Jeanotte : I should be carful when welding stainless steel. I used 93% helium as and “active gas” to increase the heat on a mig welder.

TheCarmacon : MEMS engineer here. Leaving vents in the silicon cover for HF to pass through? I doubt that, do you have a source? Instead, I wouldn't bother growing those oxide layers and simply bond a second Si wafer on top of everything, plain and simple. For resonance applications one does not simply grow oxide on a large scale and then etch it away. That results in way too much and hard to control deviations in the oscillating masses. With the HF in the cavities you'd get a lot of condensation of the reaction products. They can also detach and shift during operation, thus randomly changing the characteristics of the sensor.

Elliott S : Is there a way to use this phenomenon to measure the exact concentration of helium in any given environment?

NightHawkInLight : The enclosed oscillator is some 100+ times larger in volume than the naked one. Why is that?

Jonah Beale : 7:04 [casually roasts imperial system]

Kerry Wong : Very informative. So this means that these MEM oscillators are also sensitive to alpha radiation as the radiation produces helium. Would be interesting to see how these MEM oscs handle in a radioactive environment.

Oli K : that oscilloscope is twice as big as my future

Cadwaladr : I wonder if neon would be small enough to diffuse into the thing, seeing as it's also monoatomic like helium.

KohuGaly : What is it exactly that kills the device? Is it the helium atmosphere inside the device causing friction and viscosity and thus damping and down-tuning the oscillator? Or does the helium conduct electricity and discharges the electrodes that that should electro-statically drive the fork? BTW, the reason why there are 4 tuning forks is simple - to keep the centre of mass in single spot. When you have singe fork [ I ] , the centre of mass swings with it. When you have two [ U ] you fix the center of mass horizontally (the arms swing in and out), but there is still motion vertically (each fork moves out and down, then in and up, in and down, out and up, ...). When you put 4 forks [ X ] their vertical movement cancels out and the centre of mass stays put. The result is, that the device does not leak energy via vibrations and is less sensitive to outside vibrations.

DriveJapan : Try neon. Its the second smallest atomic radius I think. Might permeate in faster than hydrogen to see a second fail gas. Its still not near as small as helium though.

Shylesh Srinivasan : Behold our ultimate weapon against SkyNet ! Helium ! Ha Ha ! Great Video sir ! Absolutely well made ! Really grateful for this !

Vitor Pereira Tavares da Costa : Amazing video. Very interesting. I was imagining if you could do the same with the acelerometer IC, as it is also a MEMS. Would it have similar results?

Cere : Using your SEM for the benefit of other nerds. Love it! Thanks.

Kurt Hermann : It is typical in chemical processes that sorption is fast while desorption is slow. Often this difference is due to a chemical interaction between the substrate and absorbed species.

Riley Santiago : The technology used in these chips is incredible, and those electron microscope images are beautiful! Well done sir!

Dinushka M : WOW..! great video you should do a video series in semiconductor manufacturing at home.

VejyMonsta : What a strange failure scenario. Now I can go around and tell people not to put their iphone in helium, so I can explain this surprisingly complicated component failure.

Mattijs : This guy needs a youtube nobelprice or something...keeps amazing me

WGwireless : Now you have to put a known quantity of silica in a pure helium atmosphere an check how much it can adsorb, by weight.

scottdotjazzman : You are correct, that is an NW flange. I have worked in the vacuum industry as an IT support engineer for the past 8 years. Also, this was pretty fascinating; I half expected you to have a leak detector hooked up so you could see the diffusion rate once you removed the helium environment. Guess not everyone can afford a $15,000 measuring instrument that you only use once in a great while. O:-)

Michael Aichlmayr : That was outstanding detective work! You always inspire me to make better use of the things I already have, and tackle projects I thought were beyond my resources!

DrMcGordo : The "S" in MEMS should be capitalized. MicroElectroMechanical Systems. PD: great video

Ryan M. : So I’m wearing bass boosted headphones and that intro gave me goosebumps and chills

Laharl Krichevskoy : it only went up by 0.5Hz, and i was thinking that maybe we could overclock non-overclockable stuff by putting it in 1% He >_>

erezra : So why does the Helium stop the oscillator? Does it diffuse in and cause drag?

Straight Whitemale : Unreal, just unreal I am an ancient being. I remember testing vacuum tubes. I remember being amazed by transistors. We actually trimmed crystals to get the desired frequency. Thing is I am 67 years old, what will the next decade bring?

Hak Ola : this is a good discovery a weapon against AI robot in the future should we need to kill them...just poison them with Helium.....interesting

Sam Zeloof : beautifully simple and precise explanations, as always :)

Crushnaut : Two interesting questions to understand why it takes a couple days to recover. If you leave the mems chip in the helium environment for various times, does it take longer to recover? If you warm up the mems chip does it recover faster?

Gönndalf MC : Very informative video, thank you! I studied Microsystems engineering at uni. As for sealing the chamber, you mentioned a deposit, but you could also bond top it of with another wafer and then etch to the desired thickness. Its called Waferbonding, that was my first guess, it may be to expensive though.

Chemicalvamp : After the video I'm going to search "why does quartz crystal oscillate" and i hope you already have a video up :)