Hot carrier effects in real MOSFETs?

Question

My semiconductor textbook says hot carrier injection is done by applying a lot of volts (and necessarily, amps) to a FET, and electrons magically get stuck in the gate oxide. And of course, since this changes the threshold voltage permanently, it's hysteretic, so they can put them in big assed arrays with addressing and call them EPROMs. Or with another effect and another connection, EEPROMs, or in sectors, Flash.

Now that's all well and good, but is it possible to observe this in discrete FETs? Can you take a 2N7000, diode-strapped, spark a cap into it and measure a change in Vg(th)? How about IRFZ46N? What current density / terminal voltage is required to see a change? Can it even be done without burning the entire die in the process? (i.e., do EPROMs get away with it because they're burning a teensy fraction of silicon at any given time?)

Tim

Jon Slaughter · Accepted Answer

It's not permanent. The lifetime is very long but the charge can be removed.

I doubt it. Your book should have said that the gate layer is extremely thin and thinner than the standard gates used in the switches. Although I guess you can still try. It will require a much higher voltage and probably ruin the device. If I recall correctly they use two gates.

John Larkin · Answer

Eproms have floating metal or polysilicon gates that are charged by tunneling and discharged with UV. The charge is stored on the gate electrode, not in the oxide per se. The gate is designed specifically to allow all this.

I'm guessing that sustained high gate voltages will eventually shift the gate transfer curve slightly. That was a serious problem in early CMOS opamps. I think it was caused by migration of impurities in the gate oxide.

What voltage can a 2N7002 gate stand, anyhow?

John

Robert Baer · Answer

1) use no zap, frap process.  2) there is no burning in EPROMS.  3) use a curve tracer, low side to source & drain and high side to gate; carefully adjust from zero volt swing to (say) 15V maximum (value chosen from typical gate ratings).

Robert Baer · Answer

* Absolutely. And the effect was as if the input was permanently a one (if i remember correctly.   Annealing in an oven would allow those charges (both positive and negative) to dissipate and recover the use of the gate.

Hot carrier effects in real MOSFETs?

Join the Discussion

Didn't find your answer?