We have said for some long time that a silicon junction should run no hotter than 200C. If I am passing essentially microamps through the junction to use it as a cheap temperature sensor, what is the actual limitation on the temperature a silicon diode can take before destroying itself?
Any silicon device known that can beat the 1N4148 for being able to withstand heat? I suspect that the epoxy packaged 1N400x series would have some package issues, but how about the little glass diodes?
Solder and epoxy are probably bad news, so use glass s-bend terminated diodes.
I tested plastic-package power mosfets to destruction. Vgs threshold declined with temp up to 300C, at which point they were on at zero bias. They failed permanently at 320. Solder maybe?
While signal diodes aren't rated for heat (and 1N4148 is heavily multiple-sourced, so is NOT a good candidate for out-of-spec characterization), power devices ARE. A DO-41 Zener is fully characterized and rated at 200C, as is a metal-case 2N3055. Both are metal and glass construction (no plastic).
There are, I hear, some SiC diodes out there... somewhere. Anyone have experience with those at elevated temperatures?
At a third of a gram, the DO-41 might be fine for temperature sensing. It's still more than twice the mass of a DO-35
1N4148. I'm guessing the OP wants a low thermal mass and the TO-3 'can' probably won't qualify for either size or mass.
I think the "UL class 2" designated DO-35 glass packaging is specified for up to 200C operation, at least for such rated thermisters if not certain 1N4148s. Certain 1N4148s may be hard to beat on some scores if an Si junction is to be used.
There was a Unitrode guy back in the day who used to do a demo of a DO41 PIN diode--he'd run a few amps through it to heat it up, light a cigarette from it, and then let it cool down, to show that it still worked.
Dopant diffusion is probably the limiting factor if the tempcos are well matched.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
That came to mind. I also wonder about metal migration from the wire contacts. Wouldn't metals be fairly mobile and possibly poison operation at lower temperatures more readily perhaps than dopant diffusion might? I'm curious which may become more a problem first.
And there is mechanical flexure, with up and down temps. I suppose we can assume for purposes of the above question that the temperature may be hot, but stable and not oscillating wildly. So many failure modes, so little time. ;)
IIRC, and i very well may not, dopant mobility is not much of a concern below about 600 C. Band collapse causing non-function seems to be the thermal issue in comparison.
Dopant mobility seems to be possible at 250C. The reason I think so is that I worked with FABs doing low temperature processing at that temperature (I was working on visible wavelength methods of non-contact temperature measurement and these were 'difficult' temperatures for visible, narrow band work.) Ion implantation would take place, then in these unusual (then) cases, annealing at 250-300C.
But your point is also taken. Oxides were around 1500C, but they had to be quickly achieved at fast ramp rates (300C per second at the time) to avoid too much broadening of lower layers. But things survived and weren't destroyed. I haven't been doing this for years, so I'm not privy to these small feature size processes.
So I suppose it depends on the process. But it _could_ be that dopant mobility is a problem for some processed parts at
250C if annealing works as I was told it did in those special cases. Not saying it applies to the usual Si diode case, though. I think you might be right, there, because many companies worked well with optical observation starting around 400C and above. Back when.
Let me think about the "band collapse" for a moment. Thermal agitation increases the mobile charge density as it agitates more valance band electrons into equilibrium with the conduction band. I've not considered that to be a collapse of the band gap per se. Instead, it increases conductivity because of the additional charges available, though I am not convinced right now that it changes the mean free path or the drift velocity. It seems more of an effect on I_CBO, doubling every 10C. I seem to recall that Ge has a higher reference I_CBO at 25C (microamps) than Si (nanoamps) so that's more the reason why junction temps of 200C apply to Si and 100C (or so) to Ge.
Jon
P.S. Probably not right away, but I'm kind of tempted to sit down and start with the basics -- charged particles and a simple cloud model, field intensity, energy, mobility, mean free path, and drift velocity. See where that takes me with materials, then doped materials, then when they are placed into a junction. I've skimmed others' work on that but I haven't really had to do that before and it would be fun to push that rock a little bit forward on my own to gain a better personal appreciation.
Okay. So I think I understand your term. By 'band collapse' you mean that as thermal agitation increases, the equilibrium spacing between the atoms increase and that decreases the mean potential seen. Interesting, and I didn't catch that right away.
Interesting. That's pretty darn hot ! What kind was it ? Silicon FET rated for 150 C ? Or 175 C ? Something else ?
I wonder if you drive the gate negative if it would still go off at that temperature ? Maybe you had a depletion mode FET for a short while ? Probably not much of a "power" FET at that point.
TO247 power fet, rated 300 watts. I don't recall the part number.
That would be interesting, to pull the gate negative and see what happens. I was wondering about the 2N7002 drain leakage curve: would negative gate bias reduce leakage?
I acknowledge your superior experience. My rather sloppy language about bandgap reflects that. Annealing is a little bit different process, more about crystal bond reconfiguration to incorporate the implanted dopant atoms. More of a one time thing. I guess that the another useful place to look is at solid soluabilities versus temperature.
Well maybe, my language was sloppy. My thought was along the lines of = the changes in carrier population density changes versus temperature in and near the depletion region causing generalized transition from semiconductor to weak conductor. Decades since i took semiconductor physics.
That immediately came to mind -- temperature would certainly agitate a larger carrier population. The interstitial widening of atoms also lowers the charge potential as seen by other electrons -- lowers the band gap. It's interesting because at the same time the band gap is lowered for that reason, there is also an increase in the free carriers due to additional agitation, as well. Two effets. But because the band gap is lowered for the widening, still more are agitated into conduction. It's several compounding things, I guess, that result in geometrically escalating problems at higher temps.
I do enjoy thinking about these things. I'm glad for any discussion, honestly. It's nice to talk over things and it is especially nice if one can get 'kinda close' to reality from a 1st principles discussion.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.