Helium in the air kills iOS devices? (MEMS oscillator vulnerability)

[RickBrant]

Well, according to this reddit thread, yes. https://www.reddit.com/r/sysadmin/comments/9mk2o7/mri_disabled_every_ios_device_in_facility/

It appears that helium (at ordinary temperatures and at concentrations you might get during an accidental release when bringing up a new MRI scanner) devastatingly affects MEMS oscillator chips, changing their frequency enough to be out of useful range.

I think David should get some MEMS oscillators and test this!

edit: changed title to make EE(vblog) relevance more apparent.

edit: Added the following - here is a link to a paper that describes helium vulnerability in MEMS devices. It's in the reddit thread but it's buried a bit... https://file.scirp.org/pdf/JST_2013122009560886.pdf Note that ceramic and metal-packaged MEMS devices were not found to have problems.

[German_EE]

This is fascinating. It appears that the helium molecule is so small that it is able to penetrate the seals on some components that hold parts in a vacuum. One of those 'oooh, we never thought of that' moments.

[langwadt]

I'm skeptical the helium had anything to do with it, super magnets have vents to the outside so you don't suffocate people if you have a quench. The article talks about putting a mems device in 415kpa helium for days, something even remotely like that obviously didn't happen because people would die

[wraper]

I don't think helium has anything to do with this. First of all helium is naturally present in the air in small quantities. Secondly, helium is not THAT leaky that somewhat increased amount of it would penetrate vacuum seals in less than one day and destroy the device.

[thm_w]

The behavior of the devices was pretty odd. Most of them were completely dead. I plugged them in to the wall and had no indication that the device was charging.

Wouldn't an iphone have secondary independent circuit to control charging? Although it depends if there is any way to visibly see if an iphone is charging without the screen on, which I don't know.

Yes, I’m one of the employees who will be receiving a “shiny new device” tomorrow. This whole situation has me very concerned. But to correct something: our TV’s stopped working in the building as well, and a couple laptops in room directly over MRI shut down also.

So TVs and laptops were also affected, but not clear if its permanent damage or crash/reset due to EMI.

CovekIzSenke
I work in a physics laboratory and handle / work with liquid helium daily. Even during normal operation of the experiments there are moments when it evaporates to air (albeit not in huge quantities, at least not compared to a quench of a S.C. magnet). After years of exposure my phone, as well my other consumer-grade electronics, are still operating normally. I haven't heard of instances of phones or other devices malfunctioning from any of my many colleagues (in this lab or other ones). Only with a quench of the magnet could release a considerable amount of He into the atmosphere, but even then due to safety concerns I'm sure that it would be funneled to the outside to prevent suffocation, so I'm not sure that enough people would get a significant "exposure" to cause this. I'm more inclined to believe that the EM pulse from the magnet shutdown caused this, as I don't see a reason for my phone to still work after a longer and possibly higher exposures.

Interesting anecdote

luckyluke193
It would help if you could tell us what happened. I guess only the technician who was handling the system actually knows though. There are a lot of things that can go wrong with an MRI.

Contrary to what many people claim in this thread, quenching the main coil should not lead to a strong EMP by itself, since it should take on the order of seconds. The main coil is a huge inductor, so the time constant of the (quench resistor + main coil) RL circuit is pretty long. So the frequency of this EMP is of the order of 1 Hz – this is far too low to cause any serious damage across longer distances, you should think of a quasi-static field there. If this weak EMP could kill a device, it must have almost died just from being brought into the magnetic field. This does not seem to be the case.

1 litre of liquid helium turns into about 750 liters of gas at ambient conditions, so a quench produces large amounts of helium gas. While helium is chemically inert, you can asphyxiate from standing in a cloud of helium gas, and it can diffuse through some air-tight seals because a single He atom is so much smaller than any air molecule. This means that it can enter and destroy some MEMS devices that are designed for operation in vacuum. There should be a large pipe connecting the quench valve to somewhere outside the building, so most of the helium should go away. Since your technician in the MRI room did not die of asphyxiation, almost certainly no MEMS devices on the other side of the building were harmed.

Superconducting magnet power supplies are powerful devices, if they fail catastrophically, by themselves or during a magnet quench, they can cause voltage surges on the mains in your building. If this causes an EMP strong enough to kill devices through the air, it should fry everything connected to the surge via copper though. If everybody's phone died while charging via the mains, a surge would be a likely explanation.

You could get also a surge from a failing vacuum pump, for example.

The fourth possible source of problems with an MRI system is the RF system. If the system is not configured and shielded properly, and gives a random RF pulse for no reason, this can act as a legitimate EMP. What is the operating frequency of your MRI, or what is the operating field? The higher the frequency, the more efficiently it will radiate RF outwards and potentially fry things. A 1.5 T MRI uses 60 MHz, 3 T uses 120 MHz, 7 T uses 300 MHz, etc.

Based on your description, the most likely thing in my opinion seems to be that the technician did something to RF system, and it turned on the transmitter in an uncontrolled fashion in an uncontrolled hardware (and firmware/software) configuration. An EMP at your MRI operating frequency could disable sensitive electronics. This would mean that any chip could potentially be damaged. Apple might have designed their devices such that after the EMP-sensitive chip dies, the whole device immediately dies as well. This would mean that any device could be affected, the damaged chips inside just didn't kill them immediately. Test every device in every possible way to reduce the number of nasty surprises that will appear in the future. Alternatively, find a job where your systems are less likely get EMP'd randomly.

From that linked PDF, the resonator drifted by 0.2% after many hours in a high He environment. Is that enough to cause something not to even power up?

[RickBrant]

This is fascinating. It appears that the helium molecule is so small that it is able to penetrate the seals on some components that hold parts in a vacuum. One of those 'oooh, we never thought of that' moments.

They certainly should have thought of it. Containing helium is well known as a difficult problem. Helium toy balloons collapse much faster than air ones. The metallized Mylar balloons, though, hold helium much better as the metallization helps make a better barrier (and they're generally not at any significant pressure).

In the early days of HeNe laser tubes, a common failure was that the stuff would "migrate" around the glass/metal seals, leaving the wrong mix of He and Ne behind. And you could fix them by putting them in an aquarium, turned upside down, and filled with "helium" from a balloon gas bottle. STP is far higher than the needed He partial pressure in the tube, and only the He can do that trick, so it didn't matter how much other junk was in the gas. You didn't even have to remove the tube from the laser housing (assuming the unit was small enough to fit under the aquarium), and you could monitor progress by powering it up during the "refill".

[RickBrant]

I'm skeptical the helium had anything to do with it, [...]

I'm skeptical too... again, this sounds like fine material for a Dave video

[amyk]

I suspect it was an EMP event too --- if you read the Reddit comments he mentions other devices affected, and I doubt the helium would've been in sufficient concentration to have any effect (as that linked paper shows.)

[wraper]

I suspect it was an EMP event too --- if you read the Reddit comments he mentions other devices affected, and I doubt the helium would've been in sufficient concentration to have any effect (as that linked paper shows.)

Yep, it was tested at 2.8 atm pressure, thus seal experienced 2.8 more pressure than normally. If there was such helium concentration to kill the component this fast, people would have health issues or even die due to oxygen deprivation in the first place.

[BravoV]

As Hydrogen is smaller than Helium, wonder if a simple home experiment, generate H gas with simple water electrolysis, submerge the MEMS into the pure H gas chamber, and measure the drift ? Just guessing. 

[RickBrant]

As Hydrogen is smaller than Helium, wonder if a simple home experiment, generate H gas with simple water electrolysis, submerge the MEMS into the pure H gas chamber, and measure the drift ? Just guessing. 

Actually no. H atoms are larger than He.* And H always comes as H2 molecules, and it's very reactive, so it "sticks" to things. Helium comes as individual atoms and is unreactive with just about everything, so it's very "slippery".  It's far harder to contain than hydrogen.

(* http://www.batesville.k12.in.us/physics/PhyNet/Quantum/atom_quantum/answers/ch38_assignment.htm, item 30)

[SeanB]

The standard leak test is to use helium, as it pretty much will migrate through most things regarded as being good sealing materials. This is used to check semiconductors during manufacture, by allowing them to stand for a time at pressure in a helium atmosphere and then either measuring the change in mass, or checking for the gas diffusing out again. Even hermetic welded parts show some diffusion of helium through the metal and glass seals, and plastic packaged items are a lot worse than this.

[2N3055]

My electronic device stopped working next to 10 Tesla magnetic field... I think it is logical that reason for that is helium....

I'm not saying saturating devices in helium atmosphere cannot have influence..
But would start first with EMP level magnetic fields first as my first suspect..

[raptor1956]

The standard leak test is to use helium, as it pretty much will migrate through most things regarded as being good sealing materials. This is used to check semiconductors during manufacture, by allowing them to stand for a time at pressure in a helium atmosphere and then either measuring the change in mass, or checking for the gas diffusing out again. Even hermetic welded parts show some diffusion of helium through the metal and glass seals, and plastic packaged items are a lot worse than this.

Having worked in the semiconductor industry and used helium leak detectors the basic kit involves a kind of mass spec and a vacuum pump and you connect to the thing you want to leak check then you go over the thing with a wand that emits small amounts of helium.  Usually an o'ring or fitting, but sometimes the chamber itself is cracked.


Brian

[tooki]

My electronic device stopped working next to 10 Tesla magnetic field... I think it is logical that reason for that is helium....

I'm not saying saturating devices in helium atmosphere cannot have influence..
But would start first with EMP level magnetic fields first as my first suspect..

IIRC the report says that the devices started working again after a week or so. I assume EMP damage would be permanent. Whereas the resurrection after a week does make sense if the absorbed helium can dissipate, as they’re saying.

[2N3055]

My electronic device stopped working next to 10 Tesla magnetic field... I think it is logical that reason for that is helium....

I'm not saying saturating devices in helium atmosphere cannot have influence..
But would start first with EMP level magnetic fields first as my first suspect..

IIRC the report says that the devices started working again after a week or so. I assume EMP damage would be permanent. Whereas the resurrection after a week does make sense if the absorbed helium can dissipate, as they’re saying.

Report said they will get replacement devices. GE tech said it SHOULd start after a week. I didn't see any word if that happened.

[StillTrying]

Well, according to this reddit thread, yes. https://www.reddit.com/r/sysadmin/comments/9mk2o7/mri_disabled_every_ios_device_in_facility/

"Initially I thought this only impacted users on one side of the building, but from what I've heard today it seems to be multiple floors across the facility."

I think it's EMI, anyone near the source would have been speaking in a high pitched voice.

[wraper]

My electronic device stopped working next to 10 Tesla magnetic field... I think it is logical that reason for that is helium....

I'm not saying saturating devices in helium atmosphere cannot have influence..
But would start first with EMP level magnetic fields first as my first suspect..

IIRC the report says that the devices started working again after a week or so. I assume EMP damage would be permanent. Whereas the resurrection after a week does make sense if the absorbed helium can dissipate, as they’re saying.

Much more sens would be that electronics hanged in a certain way and resumed working properly once battery discharged.

[tooki]

So it conclusively is NOT EMI: the dude ran experiments using helium and was able to reproduce the problem.

https://arstechnica.com/gadgets/2018/10/helium-implicated-in-weird-iphone-malfunctions/

[Piotr Niemojewski]

https://tf.nist.gov/general/pdf/906.pdf
"The background pressure of helium inside the resonator enclosure can significantly increase if the
vacuum enclosure is glass and the resonator is operated in an environment with large amounts of
helium. A typical helium leak rate for a glass enclosure operating at 80°C in a pure heiium environment
is 5 x 10-3Pa/s."

[StillTrying]

I'm amazed that He molecules are small and light enough that a few get through the seals. but large and heavy enough to effect the MEMS.

https://ifixit.org/blog/11986/iphones-are-allergic-to-helium/

[BrianHG]

It just sounds way too sensitive... unless.... it's that they are trying to drive that 32khz reference crystal so lightly, or just at the edge of tuning, like in some old PIC32 projects I had to patch, that a fraction of a pf on the device seizes the oscillator, and once the phone has a hardwired battery, it can never be power cycled to re-kick the oscillator to run after the fact.  Something Apple has passed from phone to watch while Android devices using a different CPU circuit loading the MEMs crystal differently making them much more robust to keep oscillating.

[mikeselectricstuff]

I'm amazed that He molecules are small and light enough that a few get through the seals. but large and heavy enough to effect the MEMS.

https://ifixit.org/blog/11986/iphones-are-allergic-to-helium/

Surely the issue is that the air density is reduced, as helium displaces the air

[langwadt]

I'm amazed that He molecules are small and light enough that a few get through the seals. but large and heavy enough to effect the MEMS.

https://ifixit.org/blog/11986/iphones-are-allergic-to-helium/

Surely the issue is that the air density is reduced, as helium displaces the air

I believe they are vacuum inside, which would also explain how the helium gets in

[Wolfram]

I'm amazed that He molecules are small and light enough that a few get through the seals. but large and heavy enough to effect the MEMS.

https://ifixit.org/blog/11986/iphones-are-allergic-to-helium/

Surely the issue is that the air density is reduced, as helium displaces the air

I believe they are vacuum inside, which would also explain how the helium gets in

Diffusion only depends on the differential partial pressure. In other words, the helium will diffuse in at the same rate whether there are other gases or not. This is the reason Helium can leak out of HeNe laser tubes even if the helium pressure inside the laser tube is less than 1/100 of the air pressure on the outside of the tube.

If the packages are permeable to helium, then the helium pressure inside the package will be the same as the helium partial pressure on the outside of the package. This is around half a pascal with normal air, due to its 5 ppm helium content. An increase of helium concentration beyond 5 ppm in the surrounding air will also lead to a corresponding increase in the internal helium pressure. From the videos of iPhones being exposed to helium, it seems that this process has a time constant on the order of minutes.

As other gases are likely unable to diffuse through the MEMS oscillator package, the total internal pressure will be equivalent to the external partial pressure of helium (plus maybe hydrogen), in addition to whatever gases were left from incomplete evacuation during the manufacturing process. It therefore takes very little extra helium in the surrounding air to drastically change the pressure inside the package)

[cdev]

Why do sealed plastic bags sometimes expand with trapped gases inside over time?

[DenzilPenberthy]

Why do sealed plastic bags sometimes expand with trapped gases inside over time?

Maybe due to atmospheric pressure decreasing over that time period, or changing temperature?

[TheSteve]

A friend of mine has had his Apple watch replaced twice due to helium exposure.

edit - I should add that he does work around helium at times, the sensors are very sensitive.

[StillTrying]

"This is around half a pascal with normal air, due to its 5 ppm helium content. An increase of helium concentration beyond 5 ppm in the surrounding air will also lead to a corresponding increase in the internal helium pressure."

I know helium is strange, but I still find this MEMS problem hard to believe!

[Red Squirrel]

I bet everyone in that room was talking pretty funny too.   

[StillTrying]

Why do sealed plastic bags sometimes expand with trapped gases inside over time?

Fumes from the plastic itself ?

[BrianHG]

Tested true here:


I bet they are just driving that crystal too weakly, or just at the edge of it's tuning capacitance.

[ChunkyPastaSauce]

[ruffy91]

Tested true here:


I bet they are just driving that crystal too weakly, or just at the edge of it's tuning capacitance.

MEMS oscillators do not use quartz crystal which is piezoelectric. They use polysilicon in a electric field. They are much worse than quartz (25ppm/K temperature dependency). But there is 1 big advantage. You can make them very small and you can make them out of the same silicon as the rest of an IC, they are much smaller even including drive circuitry and temperature compensation than a quartz.
I think Apple devices are more affected because they use MEMS oscillators for all devices while most Android devices still use quartz.