Note that for a bipolar transistor, the transistor gain may not be enough to drive it from a processor pin, so you may have to use two transistors after each other
just for simplification, let's say the uC can drive 6ma - pretty close to reality. That whould require a Hfe of 1000, I recall no such transistor :)

also, since the pullup of a '51 usually is pretty miserable a PNP NPN combo should work much better.

Erik