Guess This Graph

How about

That is weird!

It's almost as if the device is behaving as a combination of a "perfect" diode, and a somewhat-temperature-sensitive resistance... but the "resistance" is somehow bypassed, or reduced in value at higher voltages and currents. I can't tell if the higher-slope part of the curve at the left is exponential or just higher-slope linear.

Perhaps there's some sort of second-order effect kicking in, to create a higher-conductivity channel at higher voltages or currents? Something analogous to avalanche in a transistor or a "hot carrier" effect, or an effect of the heating of the junction during a surge?

Interesting characteristic. The divergence above about 32.5A is interesting, too.

Needless to say, the Infineon Spice model doesn't reflect this behavior.

I guess I shall have to buy one and CT it.

Am guessing that is the resistivity of the bulk (doped) silicon.

Like maybe it would flatten out at higher voltages and currents, if you had, I suppose, a cryogenic curve?

Maybe? But avalanche is an electric field thing, even with the thinner junction in forward bias it should take quite a bit? And if the voltage drop is due to the resistive bulk, very little additional voltage will appear on the junction itself...

Heating could do it -- t_p = 200us is awfully long, and up at, say, 25A *

And with such a long pulse, when to they measure it? Do they sample at the beginning, just after series inductances have stabilized? Do they measure at the end? Do they average the whole pulse? Is their source resistive, constant current, constant voltage?

Maybe the excess heat is also activating defects and dopants? This seems to show a similar effect:

Tim

Looks like it would make quite a nice varicap diode. 1v to 10v rev bias gives a nice change in capacitance from about 140pF to 60pF.

That's nothing new -- big power schottkies will do 30nF at zero bias down to a few nF at rated voltage. They tend to stink as varactors though. That said, SiC may have lower dielectric loss!

Tim

Infineon replied!

Apparently, it's due to P-N junctions kicking in, perhaps guard ring stuff or something like that. Suffice it to say, an SiC schottky isn't a crystal with a layer of metal glued to it, they do more work to them than that.

They also gave me a reference: Heinze/Neumeister/Rupp: Surge Current Ruggedness of Silicon Carbide Schottky- and Merged-PiN-Schottky Diodes, ISPSD 2008

So if you're still curious, and have access to IEEE,

Tim