It's not concentraiting the helium; yet it's allowing it to diffuse, like the reactive non-medal is solivating a gas into it? Or is it slipping around imperfections in it's structure
Ozone will also kill all kinds of semiconductors, including LEDs. I was using an ozone generator in a cabinet and it had an LED and microcontroller both die. The concentration probably has to be fairly high and it takes days or weeks to happen.
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.
@@garymarsh23 I doubt hydrogen diffuses through silicon as a dimer. I suppose the potential well of a H interstitial in a silicon lattice is just deeper than for He
need to consider those 'constants' more carefully - keep reading and see Fig. 1.23 - molecular hydrogen diffuses faster than helium in single crystal silicon, search for articles by Douglas Sparks for more insight into this issue
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.
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.
Under normal circumstances though, these oscillators are essentially never off in the devices. They might be powered down once you run your battery down to 0 in you iPhone, but that rarely happens.
@@JasperJanssen That's probably true for the 32.768 kHz oscillator on a phone or watch as it is used for the RTC (real time clock). I still think that power cycling it would be informative, though. And on embedded systems which do not need an RTC, one generally only keeps the current oscillator on, so even the 32.768 kHz oscillator will be turned on and off to save power as the device switches to different modes, so again, interesting to test.
Embedded systems can use an small RC resonant oscillator instead of the quarts to save power, but it's much less precise. Modern CPU are too complex and timing is too strict for them to do that, they can't change de PLL that easily while running (they can change the multiplier though, to save power).
Something else to consider is that many gasses can bond to surfaces that are very clean and free of oxide. I work with high vacuum systems daily and we have problems with both helium and nitrogen bonding to the tiny wires inside our ion pumps. Usually the solution to accelerating the escape of these molecules is to heat the grid up to a very toasty 300C or so for 8 hours while constantly pumping with the turbopump. On our leak detectors we sadly do not have a way to do this, so if they become poisoned with large amounts of helium it can take days to get them to stabilize.
arfink Are you sure it’s surface bonding? Sounds like simple diffusion to me, if the oxides would form a diffusion barrier (I don’t know whether they in fact do, though).
Fragilization/embrittlement of metals (or metaloids) by H2 (or He...) induces the structural failure (SF) of the monocrystaline quartz, hence the KIA of the mem. That SF is derived from multiple dislocations (and even early microcrack developments) while in service, wich ultimately act as difussional barries during the postmortem offgassing tests. In fact, offgassing timing is an interesting way of assessing the degree of dislocation networking without resorting to TEM
@@alterhec if I follow correctly, does this mean it's possible that even the tiniest fissure (crevasse) in a material, especially an alloyed metal; would inhibit diffusion of substancially smaller particles (like elemental He) and possibly even reflect them just like how snow (transparent H2O in solid form but it looks white because of the nucleation of ice secretion) will reflect light since photons are astronomically smaller particles? Probably not the best metaphor but if you disregard the effects of Ionization, Electromagnetism, Radiopacity, etc. maybe it can be a simple model to explain the phenomenon we see with Helium. I am fascinated by this because of the implications it has on superconductivity and everything that was learned in the early days of cryogenic when using Helium to develop what was called a "quantum fluid", just thinking about how electrical conductivity is totally different at incredibly cold temperatures and how wave propagation through a material is seemingly out of this world, nonetheless diffusion of a particle being mind boggling.
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.
this kind of suggest the helium is never bonded, a part of the Si crystal structure. Then the only real force forHe to leave is random motion of the He? What if you shake the device rapidly i wonder if it leave faster?
@@marshalcraft no, Michael Kaliski is spot on with his explaination. Think of the Helium diffusing out of the device the same way you would if you flipped a full water bottle upside down, it would just glug, glug, glug. The difference in this case is that Helium is lighter than air, so it's only going to diffuse (glug) out the top of the device, but unlike the water bottle, there is nothing that can be exchanged in the vacuum. Maybe a better way of explaining this is putting your finger on the end of a straw an pulling it out of the glass. The vacuum will hold the liquid in, but not indefinitely.
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!
Thank you, your article generates some thought . From my past- Implantable Heart Pacemakers are a complex electronic device with all their sensing, programmable and telemetry. I am probably safe to say they must contain a few oscillators. Unless the process has changed greatly since I left it. What you may not know is they are baked out in a vacuum chamber followed by back filling with helium, then they are sealed in the helium dry box and then passed through an anti chamber . The next step process is they are checked in a mass spec leak detector tuned to helium, if helium shows, they are leakers. We calibrated the detector with a calibrated leak which was rated at what we told guests, was around 1cc that would take 3000 years to leak out . (Testing my memory I think the numbers were 1.6 to 10 minus 16 standard cc per second, I left there 30 years ago )
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.
That huge ceramic/metal package is actually a mems gyroscope, not an oscillator. I was just using it to show a true hermetic package. The manufacturer could have made it smaller, but it's an older device, and there wasn't as much market pressure on that product to make it tiny.
@@AppliedScience remarkable work. Can you give us a rundown of your workshop / lab setup sometimes? It would be nice to work towards being able to get the kind of insight that you're able to work out in these videos. Having a "shopping list" (or more realistically, a wish list) would be amazing to have as guidance.
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.
@@snaplash If you had access to the right kind of SEM microscope with FIB / GIS attachments it should be possible to do the required machining. Probably not simple, but almost certainly achievable. You can do some crazy things with those technologies! Look at the last "slides" of this PDF for some samples home.iitk.ac.in/~vkjain/L3-IBM-ME698.pdf
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!
This seems plausible, but is it consistent with the 1kPa environment not causing failure of the device? Perhaps it would have failed at 1kPa but it takes longer than the 24 minutes test shown?
@@GRBtutorials you should see the difference on an actual phone board. Also strongly affects the thickness of the board. Still agree that a failure mode that easy to encounter (not common, easy. There is a difference :-p) should be avoided for the cost of miniaturizing two components on a board with thousands. Imagine filling helium balloons and having your phone die for 3 days... that's just silly.
But why the failure in the first place? Does the diffused helium reduce the Q-factor of the resonant-H oscillator? (i.e. it contaminates the vacuum inside the MEMS?)
@@ChrisGJohnson It is inconceivable to me that any gas even at 1 atm should kill an oscillation. Perhaps because it is so small that surface to volume ratio so large that any gas will damp it too much. More likely the coupling is so feeble (electric field attraction only) and the desire to save power, they made the drive marginal. (I'll be willing to bet they they used @#$3.3 volts when they should have used 5.) Normally with such close clearance structures, even high gas pressure "looks like a vacuum" because there are so few molecules present in the gap. But this must not be the case. I wonder if MEMS gyros have the same problem?
So simple, yet so complicated. So much effort has been made for a device that does so little (just vibrate). The funny thing is that digital electronics can't live without that one mechanical component.
Nope,you have R C timing circuits have been done for almost the dawn of electronics. A crystal oscillator/MEMS is just a better way most of the time. Due to being much more stable.
The bit about the manufacture of a cavity inside the silicon also explains how MEMS pressure sensors are made! These things are pretty amazing - due to the stability and elasticity of the silicon cavity they can respond to pressure variations of 1/1000000 of full scale. I've personally used a 1000 bar pressure sensor as a barometer.
Thanks for the pictures and your excellent experiments and commentary, Ben! It almost looks like some kind of space ship straight out of a sci-fi novel
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.
It may also have something to do with the Reynolds number at that small scale. The effective viscosity would be really high. I saw a TH-cam video where they explained how really small animals actually swim rather than fly through air. To them, air could feel like honey.
Based on this I'm sure there are some conference rooms somewhere were someone could cause an incident by emptying out a tank of helium into the air vents.
Fascinating! Thank you for the wonderful inside look at the MEMs oscillator and especially the SEM views of the tuning forks. Your drawing also was quite marvelous.
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.
Alpha particles don't really penetrate anything, so they'd strike the outside surface, gain 2 electrons, and act like regular helium (unless it reacts with the silicon nucleus). So It would probably act just like regular helium.
I would think you would need to have a very high alpha radiation level to give the same number of nuclei as present in two percent helium gas but I imagine at some point the effect would be the same.
@@gordonwedman3179 Yes it does! As soon as the alpha particle is slowed down, it grabs up 2 electrons to make a complete helium atom. This same effect causes containers with strong alpha emitters to eventually burst because of the accumulated He gas. But alpha particles would mostly be stopped before getting in.
Many years ago I worked as a high vacuum service engineer and often we would use helium for leak detection. The vacuum system would be filled with helium above atmospheric pressure. To find the leak we used a helium detector that has a small tube sniffing the air. The tube is slowly worked around obvious potential leaks such as joints. If helium is detected then there's your leak. A more sophisticated soapy water bubble test! :)
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.
Damn Yankee That’s a different mechanism. Here it has not much to do with substrate interaction, but with partial pressure gradients across the diffusion membrane. It’s easy to produce almost arbitrarily high gradient from outside in, but going inside out, the best you can do is go down to vacuum. If the partial He pressure inside the device is very low to begin with, compared to the partial pressure available when He was forced inside, it’ll take much longer to diffuse it out. That’s the scary thing about diffusing through membranes with very high gradients pointing inside an enclosed system: it’s high impossible to get it out unless you’re willing to wait.
I should have mentioned this in the video ;) My guess is that the gas pressure inside the device causes friction between the tuning fork and the stationary electrodes, and this friction causes energy loss. If the energy loss is high enough, the oscillator will not run. It's like slowing down the pendulum of a clock with your hand. It will work with some amount of energy loss (friction), but there is a point at which it will stop due to design limits on how much energy can be put into the oscillator.
Hmm. So friction is not an issue with hydrogen, but a problem with helium? But why it doesn't happen with air? Are the major components (O2, N2, CO2, etc) of air doesn't even get into the device? Could the difference in the electrostatic properties of helium, compared to the gas already inside the MEMs device, be also a problem? Btw, thanks for the answer!
Here’s my guess. Only hydrogen and helium atoms can squeeze through the silicon, so there is never any other gasses inside. And when it’s brought out of a helium environment, the helium will start to leak back out through the silicon, but other gasses in the air cant go in, so it reverts back to a vaccum inside (besides some remaining hydrogen and helium). I guess the hydrogen is small enough or in low enough quantities that it doesn’t affect the mechanism as much as helium
Electric welding would be enough to fry any mobile device. I avoid having any mobile device near anything I am working on, since they can easily break. My grandfather wore his watch on the inside of his wrist to avoid breakage, and I never wore one at all when working, keeping a dollar store watch in my pants pocket.
Man this is super interesting! Keep these up! My jaw just drops at the ingenuity of these tiny micro devices. I would love to see more videos explaining this kind of thing.
What I really like about this video is that it answers a problem I had several years ago on a work project. We had a bar-code scanner in a fuel tank leak test chamber. Some brands/models had no problem, but the nicer ones would die very quickly. We thought it was heat buildup in the vacuum environment, (maybe 1-2 minutes) but it must have been the helium. The chamber would pull a vacuum, then pressurize the part to about 1 psi with helium. If there were any leaks, the helium would quickly be detected in the vacuum environment. The system would vent the helium into the chamber as it pumped back down (to recover the helium for the next test) and that would have exposed the bar-code scanner to a lot of helium. Some of the scanners must have had MEMs oscillators, while others had quartz oscillators.
Great research buddy! .. and that's one of the 1000's of parts that make up the cellular system, from the phone itself all the way to the satellites. Unreal amount of science, engineering, maths, ingenuity, creativity.. has come together to make this little miracle that we love so much :)
Good video :) the step where you have to "bridge the gap with another Silicon layer" is most likely done the same way as the last one, by applying filling layers (possibly assisted by lithography), then adding a new silicon layer and afterwards etching the filling layer through the holes
What a great detailed look. I'm not a regular viewer but why do you have such an amazingly well tooled lab? Let's not mention your skill in putting the tests all together.
Excellent work! I find your experiment brilliantly done. Reminds me of the engineering classes in college. We had a great time examining circuits under Electron Microscopes. Even the study of how circuit forming was done in the manufacturing process was amazing. Thank you.
Aside from the pressure argument attributed to the slow recovery (below), an additional factor that may be playing a role could be the van der Waals interaction between the Si and He. I suspect that He dissolving into the Si is not only entropically driven, but actually lowers the overall free energy, as the He still "bonds" to the Si. When you remove the He atmosphere, degassing of the Si is only entropically favorable, hence the asymmetry of failure/recovery times. Purely speculative, but thought I'd mention it. Thanks for the awesome content!
Uhhhh . . . "strange units, but at least it's KPa . . ." Uhhhh, the Pascal *IS* the metric unit of pressure . . . back up the vidjaoe, and the "strangeness" he's referring to is how the gauge measures not in absolute pressure, or even gauge pressure, as one would usually expect, but in units less than gauge pressure . . . nothing "Imperial" going on here a'tall . . .
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:-)
The amount of effort you put into your videos is unbelievable. This is high-end quality content. Keep it up and I wish you to gain everything great you deserve doing this. Thank you :)
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?
No way. Helium is smaller than hydrogen because the 2 protons suck the electrons closer to the nucleus than the 1 proton does in hydrogen. But neon has electrons in a whole different orbit than H or He.
In all your videos your ability to take measurements is outstanding. This is the hallmark of a true scientist; a skill set I envy but sorely lack. Thanks for sharing and for educating us.
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?
@Ask Questions, Try It - Apart from exposure levels, an important factor to be considered is whether the silicon of MEMS is hermetically sealed or open to gases in the atmosphere. In the case of iPhone 8 and later models, due to space constraint, the MEMS is made extremely small and is not sealed. So, if the MEMS device, - whether it is an oscillator or an accelerometer - is hermetically sealed, it will not be susceptible to helium exposure.
nice vid! I'm curious to know what would happen to other MEMS devices used in a smart phone such as microphone, accelerometer and gyroscope(not the ADI one at 4:12).
I was wondering the same thing. Does a MEMs accelerometer under constant force vibrate or is it stationary? If the former, I assume it would suffer the same problem. If the latter, it might be operating at such a low frequency that the helium has no significant effect.
Nope. It's easy to build a divider for binary powers, so the 32768 Hz 'standard' was a natural choice for the wristwatch industry and other users that need to derive a 1-second tick from the oscillator output.
It's a traditional watch crystal frequency because of the reasons NoahFect mentioned. It's still used for keeping real-time clocks up to date. I wonder though if the MEMS device in question actually works at this frequency, because it's fairly low for this kind of voodoo. Probably the frequency of the MEMS is several orders higher and the output is divided by the helper circuitry on the other chip.
@@svofski - You are right. The 32.768KHz frequency is low and is used only for time keeping. The frequency of the main oscillator in a cell phone would be 16 or 32MHz. Ben does mention that at 2:42.
@@svofski - You are right. The 32.768KHz frequency is low and is used only for time keeping. The frequency of the main oscillator in a cell phone would be 16 or 32MHz. Ben does mention that at 2:42.
@@svofski - You are right. The 32.768KHz frequency is low and is used only for time keeping. The frequency of the main oscillator in a cell phone would be 16 or 32MHz. Ben does mention that at 2:42.
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.
If the MEMS 'tuning fork' is made to work in a vacuum then the presence of any gas will cause a huge amount of damping. It may even cause current to flow between parts that are supposed to be insulated resulting in no force generation or shorting of the pick-up signal to the common.
The gas can't escape because of the built in check valve by nature...really....really not a clue, just an obtuse comment from a guy that can't wrap his head around how all this minuscule stuff works in the first place but am fascinated by it! Thanks very much for taking the time to produce these videos. But then on the other hand I feel really dumb now......
According to the articles, the failures happened while an MRI machine was installed. During that installation, two critical things happen that can lead to a lot of He gas being emitted: 1. The superconducting magnet coil needs to be cooled to liquid He temperatures in order to actually become superconducting. To do that, you essentially pour liquid He onto the coil until it is cold enough, and in the process of course a lot of He evaporates. 2. The coil needs to be energized. At least for NMR magnets (I assume the process for MRI magnets is very similar), the way the current is put into the coil is by placing a charging rod into the cryostat which connects the leads from the power supply used for energizing to the actual superconducting coil. The charging rod with all its wiring inside has a fairly large thermal conductivity, so the He boiloff is higher during energizing than during normal operation because of that. In addition, a small segment of the superconducting coil (the segment between the terminals) needs to be heated above its critical temperature during the whole energizing process to be able to charge at all, and this heat is, you guessed it, also carried away by evaporated He. A third possibility is that the magnet quenches (loses its superconductivity) during energizing, which basically always leads to a complete evaporation of the liquid He inventory of the magnet (the energy already in the coil is dumped into the now resistive coil material as heat and carried away by the He). So during installation, He levels can easily reach a few percent in a small enough magnet room. In normal operation, the He boiloff should be minimal.
holy crap that's interesting.
Hi, Nurdrage ! Didn't expect to see you wandering outside your lab.
Weird phenomenon, eh?
Why does helium diffuse into silicon while hydrogen does not?
It's not concentraiting the helium; yet it's allowing it to diffuse, like the reactive non-medal is solivating a gas into it? Or is it slipping around imperfections in it's structure
Ozone will also kill all kinds of semiconductors, including LEDs. I was using an ozone generator in a cabinet and it had an LED and microcontroller both die. The concentration probably has to be fairly high and it takes days or weeks to happen.
@@TheLightningStalker Ozone is an oxidizer, so I am sure it will kill sensitive things, but by totally different means.
Thanks, very interesting!
Dudeeeee this is like a crossover event I never even realized happened! Love both of your guy's content.
thank you for showing conclusions first and I presented the data as if it was a mystery novel.
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.
My question is why? Helium is literally twice the size of hydrogen... how does it diffuse almost 100x faster?
@@tomewyrmdraconus837 A H2 molecule is a lot larger than a He atom.
@@garymarsh23 I doubt hydrogen diffuses through silicon as a dimer. I suppose the potential well of a H interstitial in a silicon lattice is just deeper than for He
@@tomewyrmdraconus837 He_1(atomic) is smaller than H_2 (molecule) - the real question is - why would Sulphur diffuse so rapidly?
need to consider those 'constants' more carefully - keep reading and see Fig. 1.23 - molecular hydrogen diffuses faster than helium in single crystal silicon, search for articles by Douglas Sparks for more insight into this issue
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.
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.
Yes. Power-cycling during the He exposure is a great idea!
Under normal circumstances though, these oscillators are essentially never off in the devices. They might be powered down once you run your battery down to 0 in you iPhone, but that rarely happens.
I had a car like that :)
@@JasperJanssen That's probably true for the 32.768 kHz oscillator on a phone or watch as it is used for the RTC (real time clock). I still think that power cycling it would be informative, though.
And on embedded systems which do not need an RTC, one generally only keeps the current oscillator on, so even the 32.768 kHz oscillator will be turned on and off to save power as the device switches to different modes, so again, interesting to test.
Embedded systems can use an small RC resonant oscillator instead of the quarts to save power, but it's much less precise. Modern CPU are too complex and timing is too strict for them to do that, they can't change de PLL that easily while running (they can change the multiplier though, to save power).
Wow! Very interesting. Thank you for this information and your investigations.
Have something you might want to research. Health effects of microwaves at 5 gigahertz. How juxtaposition effects cells and body.
@@stevepence9869 it's depends on power and distance from the source. Are you interesting wifi or military radars?
@@AxGxP WiFi.
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!
100%
Interesting! I work in the semiconductor industry, and we use helium to leak check systems under a vacuum. Good to know!
You can probably check to see what oscillator your phone uses, but unless it's an iPhone 8 (or newer), you're probably safe.
Yes and btw, have to be careful about permeation when doing that since it will pass through o-rings etc. and indicate a false leak!
Something else to consider is that many gasses can bond to surfaces that are very clean and free of oxide. I work with high vacuum systems daily and we have problems with both helium and nitrogen bonding to the tiny wires inside our ion pumps. Usually the solution to accelerating the escape of these molecules is to heat the grid up to a very toasty 300C or so for 8 hours while constantly pumping with the turbopump. On our leak detectors we sadly do not have a way to do this, so if they become poisoned with large amounts of helium it can take days to get them to stabilize.
arfink Are you sure it’s surface bonding? Sounds like simple diffusion to me, if the oxides would form a diffusion barrier (I don’t know whether they in fact do, though).
Fragilization/embrittlement of metals (or metaloids) by H2 (or He...) induces the structural failure (SF) of the monocrystaline quartz, hence the KIA of the mem.
That SF is derived from multiple dislocations (and even early microcrack developments) while in service, wich ultimately act as difussional barries during the postmortem offgassing tests.
In fact, offgassing timing is an interesting way of assessing the degree of dislocation networking without resorting to TEM
@@alterhec if I follow correctly, does this mean it's possible that even the tiniest fissure (crevasse) in a material, especially an alloyed metal; would inhibit diffusion of substancially smaller particles (like elemental He) and possibly even reflect them just like how snow (transparent H2O in solid form but it looks white because of the nucleation of ice secretion) will reflect light since photons are astronomically smaller particles?
Probably not the best metaphor but if you disregard the effects of Ionization, Electromagnetism, Radiopacity, etc. maybe it can be a simple model to explain the phenomenon we see with Helium.
I am fascinated by this because of the implications it has on superconductivity and everything that was learned in the early days of cryogenic when using Helium to develop what was called a "quantum fluid", just thinking about how electrical conductivity is totally different at incredibly cold temperatures and how wave propagation through a material is seemingly out of this world, nonetheless diffusion of a particle being mind boggling.
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.
this kind of suggest the helium is never bonded, a part of the Si crystal structure. Then the only real force forHe to leave is random motion of the He? What if you shake the device rapidly i wonder if it leave faster?
@@marshalcraft no, Michael Kaliski is spot on with his explaination. Think of the Helium diffusing out of the device the same way you would if you flipped a full water bottle upside down, it would just glug, glug, glug. The difference in this case is that Helium is lighter than air, so it's only going to diffuse (glug) out the top of the device, but unlike the water bottle, there is nothing that can be exchanged in the vacuum. Maybe a better way of explaining this is putting your finger on the end of a straw an pulling it out of the glass. The vacuum will hold the liquid in, but not indefinitely.
Fascinating Ben!
Fascinating! Thanks for enlightening me. Science is the best!
Omg Jesus hellos 👋
Achievement unlocked
Jesus enlightened
@@Mg3-Si2-O5-OH4 bruh
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!
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...😂😂
Except that would only work with iPhone 8 or later... And it'd probably cause more problems than would solve.
Maybe we need to send mass shipments of helium balloons to all the Apple Stores?
@@JlerchTampa Or empty a few bottles of Helium into some Apple warehouses …
They couldn't put in a few drops of epoxy to isolate it?
I wonder if there is any permanent damage even after the He has diffused back out of the device...
Thank you, your article generates some thought . From my past- Implantable Heart Pacemakers are a complex electronic device with all their sensing, programmable and telemetry. I am probably safe to say they must contain a few oscillators. Unless the process has changed greatly since I left it. What you may not know is they are baked out in a vacuum chamber followed by back filling with helium, then they are sealed in the helium dry box and then passed through an anti chamber . The next step process is they are checked in a mass spec leak detector tuned to helium, if helium shows, they are leakers. We calibrated the detector with a calibrated leak which was rated at what we told guests, was around 1cc that would take 3000 years to leak out . (Testing my memory I think the numbers were 1.6 to 10 minus 16 standard cc per second, I left there 30 years ago )
The technology used in these chips is incredible, and those electron microscope images are beautiful! Well done sir!
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.
beautifully simple and precise explanations, as always :)
Another great video!
The enclosed oscillator is some 100+ times larger in volume than the naked one. Why is that?
That huge ceramic/metal package is actually a mems gyroscope, not an oscillator. I was just using it to show a true hermetic package. The manufacturer could have made it smaller, but it's an older device, and there wasn't as much market pressure on that product to make it tiny.
@@AppliedScience remarkable work. Can you give us a rundown of your workshop / lab setup sometimes? It would be nice to work towards being able to get the kind of insight that you're able to work out in these videos. Having a "shopping list" (or more realistically, a wish list) would be amazing to have as guidance.
128 MEMs-engineers and some from apple disliked this video for sure!
Thank you Applied Science for this really informative and interesting video! :)
when you see this ratio your have to realize that a good portion of thumbs downs are accidental.
Would be nice to see the dislike counts, but TH-cam just rolled out the removal to all videos...
This is probably the best coverage/video I've seen on this topic. Thanks for the upload!
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.
If you open a small hole in the MEMS, you probably will get Air in it too, which on its own would probably enough to kill the oscillator
@@snaplash If you had access to the right kind of SEM microscope with FIB / GIS attachments it should be possible to do the required machining. Probably not simple, but almost certainly achievable. You can do some crazy things with those technologies! Look at the last "slides" of this PDF for some samples home.iitk.ac.in/~vkjain/L3-IBM-ME698.pdf
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!
This seems plausible, but is it consistent with the 1kPa environment not causing failure of the device?
Perhaps it would have failed at 1kPa but it takes longer than the 24 minutes test shown?
Yeah, you don't really gain that much space with that MEMS oscillator.
@@GRBtutorials you should see the difference on an actual phone board. Also strongly affects the thickness of the board.
Still agree that a failure mode that easy to encounter (not common, easy. There is a difference :-p) should be avoided for the cost of miniaturizing two components on a board with thousands. Imagine filling helium balloons and having your phone die for 3 days... that's just silly.
But why the failure in the first place? Does the diffused helium reduce the Q-factor of the resonant-H oscillator? (i.e. it contaminates the vacuum inside the MEMS?)
@@ChrisGJohnson It is inconceivable to me that any gas even at 1 atm should kill an oscillation. Perhaps because it is so small that surface to volume ratio so large that any gas will damp it too much. More likely the coupling is so feeble (electric field attraction only) and the desire to save power, they made the drive marginal. (I'll be willing to bet they they used @#$3.3 volts when they should have used 5.) Normally with such close clearance structures, even high gas pressure "looks like a vacuum" because there are so few molecules present in the gap. But this must not be the case.
I wonder if MEMS gyros have the same problem?
So simple, yet so complicated. So much effort has been made for a device that does so little (just vibrate).
The funny thing is that digital electronics can't live without that one mechanical component.
Nope,you have R C timing circuits have been done for almost the dawn of electronics. A crystal oscillator/MEMS is just a better way most of the time.
Due to being much more stable.
The bit about the manufacture of a cavity inside the silicon also explains how MEMS pressure sensors are made! These things are pretty amazing - due to the stability and elasticity of the silicon cavity they can respond to pressure variations of 1/1000000 of full scale. I've personally used a 1000 bar pressure sensor as a barometer.
"At least its kpa" lol
None of that rubbish PSI...
@@invendelirium I doubt it as he says "strange units".
could have been cm of h2o
kilo pascals
@@gordonwedman3179 imperial>metric
I think i have subscribed over 50 channels, but you are my most favorite one! Just love how casual you investigate the most scientific topics.
the geometry of that mems is blowing my mind. Is there anyway to view the electron microscope images in higher resolution?
I updated the description with a Google Drive link: drive.google.com/drive/folders/1l3mJ4UTs8aY70scH7vDaf0M8pLeP2kqI?usp=sharing
@@AppliedScience thanks!
Thanks for the pictures and your excellent experiments and commentary, Ben! It almost looks like some kind of space ship straight out of a sci-fi novel
@@Micah561 Try shouting "Magnify!" at your screen.
@@spankeyfish
*COMPUTER!* (bleep)
Magnify image. (bleep)
Enhance quality. (bleep)
Thank you for the thoughtful and rigorous experimentation. Love that your inquisitiveness led to an answer we can all appreciate!
WOW, that's really amazing! I had no idea manufacturing could be that precise. I'm really blown away.
What an awesome tear-down/investigation! Love the level of detail, ego-less inquiry & yet in language that most people can understand. Thanks so much.
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.
I’d guess its because the penetrating helium is causing strain on the silicon and making its oscillations change?
you mean like, the presence of helium atoms in the crystal lattice changes the hardness/flexibility of the silicon?
KohuGaly or it's just friction between "moving" parts and helium
It may also have something to do with the Reynolds number at that small scale. The effective viscosity would be really high. I saw a TH-cam video where they explained how really small animals actually swim rather than fly through air. To them, air could feel like honey.
Based on this I'm sure there are some conference rooms somewhere were someone could cause an incident by emptying out a tank of helium into the air vents.
that oscilloscope is twice as big as my future
...also twice as bright!
@@realedna Danged millennials, walking around like they rent the place.
Fascinating! Thank you for the wonderful inside look at the MEMs oscillator and especially the SEM views of the tuning forks. Your drawing also was quite marvelous.
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.
I think alpha radiation consists of a helium nucleus. I do not believe alpha radiation creates helium.
@@gordonwedman3179 You're right but it's still Helium, since it's number of protons don't change. Alpha Radiation is just a Helium cation.
Alpha particles don't really penetrate anything, so they'd strike the outside surface, gain 2 electrons, and act like regular helium (unless it reacts with the silicon nucleus). So It would probably act just like regular helium.
I would think you would need to have a very high alpha radiation level to give the same number of nuclei as present in two percent helium gas but I imagine at some point the effect would be the same.
@@gordonwedman3179 Yes it does! As soon as the alpha particle is slowed down, it grabs up 2 electrons to make a complete helium atom. This same effect causes containers with strong alpha emitters to eventually burst because of the accumulated He gas. But alpha particles would mostly be stopped before getting in.
I saw your tweet about this a while back, but I had no idea you were working on a video about it! Awesome!
Using your SEM for the benefit of other nerds. Love it! Thanks.
Many years ago I worked as a high vacuum service engineer and often we would use helium for leak detection. The vacuum system would be filled with helium above atmospheric pressure. To find the leak we used a helium detector that has a small tube sniffing the air. The tube is slowly worked around obvious potential leaks such as joints. If helium is detected then there's your leak. A more sophisticated soapy water bubble test! :)
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.
Damn Yankee That’s a different mechanism. Here it has not much to do with substrate interaction, but with partial pressure gradients across the diffusion membrane. It’s easy to produce almost arbitrarily high gradient from outside in, but going inside out, the best you can do is go down to vacuum. If the partial He pressure inside the device is very low to begin with, compared to the partial pressure available when He was forced inside, it’ll take much longer to diffuse it out. That’s the scary thing about diffusing through membranes with very high gradients pointing inside an enclosed system: it’s high impossible to get it out unless you’re willing to wait.
@@absurdengineering It forms a chemical compound called silicon heliide
I learn something new every time I watch one of your videos. Thanks for the quality content!!!
One of my favorite videos of yours. Every video is such a treat!
I was just reading up on this today and I really appreciate the more in depth analysis you provided!
But I don't understand what is the mechanism how the helium makes the MEMs device to fail but not the hidrogen. :(
I should have mentioned this in the video ;) My guess is that the gas pressure inside the device causes friction between the tuning fork and the stationary electrodes, and this friction causes energy loss. If the energy loss is high enough, the oscillator will not run. It's like slowing down the pendulum of a clock with your hand. It will work with some amount of energy loss (friction), but there is a point at which it will stop due to design limits on how much energy can be put into the oscillator.
Hmm. So friction is not an issue with hydrogen, but a problem with helium? But why it doesn't happen with air? Are the major components (O2, N2, CO2, etc) of air doesn't even get into the device? Could the difference in the electrostatic properties of helium, compared to the gas already inside the MEMs device, be also a problem? Btw, thanks for the answer!
Here’s my guess. Only hydrogen and helium atoms can squeeze through the silicon, so there is never any other gasses inside. And when it’s brought out of a helium environment, the helium will start to leak back out through the silicon, but other gasses in the air cant go in, so it reverts back to a vaccum inside (besides some remaining hydrogen and helium). I guess the hydrogen is small enough or in low enough quantities that it doesn’t affect the mechanism as much as helium
Helium is a smaller molecule than hydrogen, so diffusion is much faster
Hydrogen gas exists mostly as H2 where helium is single He atoms. So even though a hydrogen atom is smaller, in gas form it's not.
You’re videos are so awesome! Thank you so much for the amazing tests and details!!!
I should be carful when welding stainless steel. I used 93% helium as and “active gas” to increase the heat on a mig welder.
Carful? You mean "as much or as many as a car will hold"?
@@GRBtutorials Good eye! I can't believe that slipped past my spelling detector.
Electric welding would be enough to fry any mobile device. I avoid having any mobile device near anything I am working on, since they can easily break. My grandfather wore his watch on the inside of his wrist to avoid breakage, and I never wore one at all when working, keeping a dollar store watch in my pants pocket.
The quality of your videos is just incredible!
I think the technology used to make the buried empty space is called "Silicon On Nothing" or SON
Man this is super interesting! Keep these up! My jaw just drops at the ingenuity of these tiny micro devices. I would love to see more videos explaining this kind of thing.
What I really like about this video is that it answers a problem I had several years ago on a work project. We had a bar-code scanner in a fuel tank leak test chamber. Some brands/models had no problem, but the nicer ones would die very quickly. We thought it was heat buildup in the vacuum environment, (maybe 1-2 minutes) but it must have been the helium. The chamber would pull a vacuum, then pressurize the part to about 1 psi with helium. If there were any leaks, the helium would quickly be detected in the vacuum environment. The system would vent the helium into the chamber as it pumped back down (to recover the helium for the next test) and that would have exposed the bar-code scanner to a lot of helium. Some of the scanners must have had MEMs oscillators, while others had quartz oscillators.
Me: *puts friend’s iPhone in bag and pops balloon into it*
Friend: Ha! I know what you’re doing - it won’t float!
Me: Oh, you just wait >:)
Great research buddy!
.. and that's one of the 1000's of parts that make up the cellular system, from the phone itself all the way to the satellites. Unreal amount of science, engineering, maths, ingenuity, creativity.. has come together to make this little miracle that we love so much :)
MicroElectroMechanical --- Thanks Ben!
Good video :) the step where you have to "bridge the gap with another Silicon layer" is most likely done the same way as the last one, by applying filling layers (possibly assisted by lithography), then adding a new silicon layer and afterwards etching the filling layer through the holes
"HF vapor"
* shudders *
What a great detailed look. I'm not a regular viewer but why do you have such an amazingly well tooled lab? Let's not mention your skill in putting the tests all together.
WOW..! great video
you should do a video series in semiconductor manufacturing at home.
thumbs up so he will see it
should produce small ics
i second this notion, as basically most important topic
Excellent work! I find your experiment brilliantly done. Reminds me of the engineering classes in college. We had a great time examining circuits under Electron Microscopes. Even the study of how circuit forming was done in the manufacturing process was amazing. Thank you.
So these are tiny helium sensors.
Aside from the pressure argument attributed to the slow recovery (below), an additional factor that may be playing a role could be the van der Waals interaction between the Si and He. I suspect that He dissolving into the Si is not only entropically driven, but actually lowers the overall free energy, as the He still "bonds" to the Si. When you remove the He atmosphere, degassing of the Si is only entropically favorable, hence the asymmetry of failure/recovery times. Purely speculative, but thought I'd mention it. Thanks for the awesome content!
The "S" in MEMS should be capitalized. MicroElectroMechanical Systems.
PD: great video
nobody cares
Certainly you don't.
this raises the question of if conformal coating over the MEMs oscillator would be good enough to shield it from helium poisoning.
7:04 [casually roasts imperial system]
Uhhhh . . . "strange units, but at least it's KPa . . ." Uhhhh, the Pascal *IS* the metric unit of pressure . . . back up the vidjaoe, and the "strangeness" he's referring to is how the gauge measures not in absolute pressure, or even gauge pressure, as one would usually expect, but in units less than gauge pressure . . . nothing "Imperial" going on here a'tall . . .
@@Roonasaur He meant that even the strange pressure measurement of the gauge is easier to work with than the imperial system
@@MegaFPVFlyer Ok, sure . . . I'm American, so I guess my worldview doesn't revolve around how weird it is.
@@Roonasaur woosh
@@RainBoxRed Yeah yeah yeah, I earned that one I guess. It's YT - I frequently fire off here half-cocked lol
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:-)
So can you overclock your smartphone with a 0,05% atmosphere Helium booster shot?
Yes, you'll get a 0.00152587890625% increase of your phone performance. But it'll wear off within days and you'll have to re-apply it.
Incredible work, I had no idea these even existed!
The amount of effort you put into your videos is unbelievable. This is high-end quality content. Keep it up and I wish you to gain everything great you deserve doing this. Thank you :)
Applied Science: Thank you for so many interesting and useful experiments and demonstrations. You are a very bright and talented guy.
I wonder how many balloons would make up 2% of an Apple store?
Don't bother with balloons. Just buy a small helium cylinder.
Outstanding. Both your home shop science and the analysis of the underlying technology. Thank you.
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?
I'm older than you. I remember when we used tin cans connected by strings for telephones!
TropicalCoder Those phones had a weird dial tone!
Wow dude, you did it again. Very cool. And I'm glad you got your electron microscope working properly.
Thanks,.
I wonder if neon would be small enough to diffuse into the thing, seeing as it's also monoatomic like helium.
No way. Helium is smaller than hydrogen because the 2 protons suck the electrons closer to the nucleus than the 1 proton does in hydrogen. But neon has electrons in a whole different orbit than H or He.
In all your videos your ability to take measurements is outstanding. This is the hallmark of a true scientist; a skill set I envy but sorely lack. Thanks for sharing and for educating us.
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?
My understanding is all mems devices are susceptible to helium exposure, but exposure to levels of helium that cause problems are not common.
@Ask Questions, Try It - Apart from exposure levels, an important factor to be considered is whether the silicon of MEMS is hermetically sealed or open to gases in the atmosphere. In the case of iPhone 8 and later models, due to space constraint, the MEMS is made extremely small and is not sealed. So, if the MEMS device, - whether it is an oscillator or an accelerometer - is hermetically sealed, it will not be susceptible to helium exposure.
Crazy amount of work in this 20m video. Thanks for the insight!
Applied Science == like
Assert.assertTrue(AppliedScience == like) ; //very true!
True
#include "youtube.h"
int main(int argc, char* argv[]) {
if (argv[1] == "Applied Science") {
like();
puts("Liked");
}
return 0;
}
Fascinating! Thank you taking the time to investigate and share this :) subscribed!
nice vid! I'm curious to know what would happen to other MEMS devices used in a smart phone such as microphone, accelerometer and gyroscope(not the ADI one at 4:12).
I was wondering the same thing. Does a MEMs accelerometer under constant force vibrate or is it stationary? If the former, I assume it would suffer the same problem. If the latter, it might be operating at such a low frequency that the helium has no significant effect.
I'm a first time viewer and gotta admit, rarely have I seen a channel name as appropriate as this one.
Is there a way to use this phenomenon to measure the exact concentration of helium in any given environment?
There are similar sensors,
As a sensor this would be too slow, but I´m sure you could make a sensor in a similar way.
take some Iphones and avarage the times they need to stop working. I am sure you can correlate this value with the amount of he in the air^^
This video just made me appreciate electronics and chemistry much more. I'm so exited, great contribution!
That 8-channel oscilloscope looks like it cost at least 500000$ :D
@No Idol 60GHz is crazy, damn!
Always so crazy what you are able to do by your own in your own shop. Always really really cool, thank you for all the effort :-)
Is it a coincidence that the normal resonance frequency is approx. 32768 Hz, which amounts to 2^15?
Nope. It's easy to build a divider for binary powers, so the 32768 Hz 'standard' was a natural choice for the wristwatch industry and other users that need to derive a 1-second tick from the oscillator output.
It's a traditional watch crystal frequency because of the reasons NoahFect mentioned. It's still used for keeping real-time clocks up to date. I wonder though if the MEMS device in question actually works at this frequency, because it's fairly low for this kind of voodoo. Probably the frequency of the MEMS is several orders higher and the output is divided by the helper circuitry on the other chip.
@@svofski - You are right. The 32.768KHz frequency is low and is used only for time keeping. The frequency of the main oscillator in a cell phone would be 16 or 32MHz. Ben does mention that at 2:42.
@@svofski - You are right. The 32.768KHz frequency is low and is used only for time keeping. The frequency of the main oscillator in a cell phone would be 16 or 32MHz. Ben does mention that at 2:42.
@@svofski - You are right. The 32.768KHz frequency is low and is used only for time keeping. The frequency of the main oscillator in a cell phone would be 16 or 32MHz. Ben does mention that at 2:42.
so excited when i see you posted a video
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.
Thanks Ben for (again) a fascinating video.
So why does the Helium stop the oscillator? Does it diffuse in and cause drag?
It changes the physical parameters of the crystal.not really drag as you aren't moving crystal dislocations around
If the MEMS 'tuning fork' is made to work in a vacuum then the presence of any gas will cause a huge amount of damping. It may even cause current to flow between parts that are supposed to be insulated resulting in no force generation or shorting of the pick-up signal to the common.
@@KallePihlajasaari well, some damping anyways. Most MEMS are small, but not That small. And these devices are designed to run in atmosphere.
@@RobertSzasz MEMS oscillators will always have a vacuum inside, the little "tuning fork" would not be able to oscillate in atmosphere
Fascinating! Nice SEM prep - quick and effective!
Ne is only slightly larger; got any lying around?
The gas can't escape because of the built in check valve by nature...really....really not a clue, just an obtuse comment from a guy that can't wrap his head around how all this minuscule stuff works in the first place but am fascinated by it! Thanks very much for taking the time to produce these videos. But then on the other hand I feel really dumb now......
Do you post your electron microscope images anywhere?
I added it just now: drive.google.com/drive/folders/1l3mJ4UTs8aY70scH7vDaf0M8pLeP2kqI?usp=sharing
@@AppliedScience Thank you i find that image of the mems device just amazing
Your videos are Always fun to Watch and Very interesting! Keep up that great Work :)
So does this break, say, MEMS compasses too?
It should, I believe all mems devices are considered susceptible to alterations by helium exposure.
It would depend on the packaging, too (some are encased in ceramic which may have different characteristics)
According to the articles, the failures happened while an MRI machine was installed. During that installation, two critical things happen that can lead to a lot of He gas being emitted:
1. The superconducting magnet coil needs to be cooled to liquid He temperatures in order to actually become superconducting. To do that, you essentially pour liquid He onto the coil until it is cold enough, and in the process of course a lot of He evaporates.
2. The coil needs to be energized. At least for NMR magnets (I assume the process for MRI magnets is very similar), the way the current is put into the coil is by placing a charging rod into the cryostat which connects the leads from the power supply used for energizing to the actual superconducting coil. The charging rod with all its wiring inside has a fairly large thermal conductivity, so the He boiloff is higher during energizing than during normal operation because of that.
In addition, a small segment of the superconducting coil (the segment between the terminals) needs to be heated above its critical temperature during the whole energizing process to be able to charge at all, and this heat is, you guessed it, also carried away by evaporated He.
A third possibility is that the magnet quenches (loses its superconductivity) during energizing, which basically always leads to a complete evaporation of the liquid He inventory of the magnet (the energy already in the coil is dumped into the now resistive coil material as heat and carried away by the He).
So during installation, He levels can easily reach a few percent in a small enough magnet room. In normal operation, the He boiloff should be minimal.
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 >_>
I was literally looking for sem images of these oscillators after reading that article yesterday. Good timing XD
A normal tuning fork would be higher frequency in helium (or try to be), so it makes sense why this failed
It would be higher frequency than when running in air, but that little cavity is supposed to be a vacuum, so helium should still slow it down.