I came up with this doubt last week. Consider a NPN transistor. Usually we bias it so that current enters through the collector and base, and leaves through the emitter, right? That allows the base to form some holes, which recombine with the electrons emitted by the emissor and that way the bugger works. Now, say I set my my circuit like this: I set a current source of 2 mA that goes to a 10kΩ resistor which then goes to the collector. Then I set a current drain that pulls 1 mA from a 10kΩ resistor which is connected to the emissor. And I set the base on the ground. So Vcb > 0, and Vbe > 0, so the thingie is in forward active mode. But Ic = 2 mA and Ie = 1 mA, so Ib must be -1 mA... the current is flowing out of of the base. If the current flows out of the base, then there are no holes generated there, and the transistor should cut off... but it can't cut off because it's in forward active mode, and also because there is nothing in the general Ebers-Moll model that actually prevents Ib from being negative... How to resolve this dilemma?