Can a transistor be damaged by limited reverse Vbe current?

[SilverSolder]

Picture a run-of-the-mill small signal NPN transistor, with the emitter tied to ground and collector to a positive supply. The base is connected to -15V via a 10K resistor. 

This would cause the base-emitter junction to enter into avalanche mode and look like a zener diode, at some constant voltage (6V - 10V or so).

Can the transistor conceivably be damaged by this small current constantly, year in and year out, for several decades?

[langwadt]

afaik even if it doesn't immediately damages it, avalanching the base-emitter will make the beta of the transistor drop over time.

story goes that it was a common fault with reversed transistors used as low drop switches and that you could recover some of the
lost gain by heating the transistor

[David Hess]

Base-emitter breakdown will lower the hfe (beta) resulting in higher noise and higher bias current.  This does not matter so much with switching transistors used in circuits with sufficient forced beta, gold doped transistors (RF and fast saturated switching transistors) which have low hfe anyway, and power transistors which operate at high currents where gain drops.  But it is fatal for precision and low input bias current circuits like differential pairs.

I think the damage mechanism is hot carriers causing dislocations in the crystal structure which lowers minority carrier lifetime.  The gold doping used in fast saturated switching and RF transistors also works by lowering minority carrier lifetime explaining why these transistors are less effected.  Radiation damage also occurs through this mechanism.

The transistor junction can be repaired through annealing but that is hardly feasible with a plastic package.

[T3sl4co1l]

Incidentally, every so often you see higher voltage diodes advertising similar methods (electron bombardment, transition metal (usually Pt) doping) to control carrier lifetime.  At very light doping and over a gently graded junction, this gives reasonable performance (t_rr ~ 30ns) for 600V diodes, at modest expense to Vf (say, Vf = 1-3V at rated current).

And it seems to not work so well at higher voltages (~200ns for 1200V, and I think I've seen 2-4us "fast recovery" hockeypucks in the 3-6kV range, which is about as high as Si can possibly go), which makes sense as carrier lifetime depends on doping, and with lighter doping, yeah.

Which, by the way, is a good reason why you may find RF or switching or power types stink at low currents -- hFE drops off quickly.  Some power transistors this is explicit -- Darlingtons usually, with internal B-E resistors.  This doesn't break Ebers-Moll or whatever, so you still get the transconductance you expect, but the base conductance can be higher than you might expect.

Tim