It is worth noting that a BJT base metallization
is liable to be much narrower than emitter
(the collector, is the whole metallized backside
of the die) and the E-B breakdown path is much
less current-capable than the expected C-E
current path (betting on beta-active or at
least forced-beta-saturated (~ 10) factor of
expected current @ max rated Ic, and layout
would have divided the available die area by
about that ratio as allocated to base and emitter
metal. So upshot is, you might encounter a fuse
current in the 10-100mA range rather than the
100mA - 1A Ic(max) the device might advertise.
A designed zener diode would have (a) the
expectation of equal and rated terminal currents
and (b) often built as a mesa diode with full face
metallization (no routing at all, just a slab)
and the leads coined directly to the die for
a much higher current withstand. The tri-metal
endcaps and large vertical extent will preclude
the aluminum-spiking "zener zap programming"
which can also be an outcome of pushing too
much reverse current on an aluminum-interconnect
junction, if you heat the contact region to the Si-Al
eutectic you will entrain aluminum and short the
junction. Though it's probably a foot-race between
interconnect fusing and junction "programming"
that depends on the E, B contacts' proximity and
layout-induced current crowding (optimized for
"easy damage" for a zap-trim element, the
opposite for a should-be-rugged discrete BJT).
It's difficult to know from the usual docs, what the
frontside metal stack of a discrete BJT actually is.
Old timey types stand a good chance of being
aluminum-only, and even high power ones may
use Al-Ti-Ni-Au (which could still deliver the
aluminum) or Ti-Ni-Au (which would form a
Ti-silicide barrier and eliminate this specific
failure mode).