The TIP120 is an NPN Darlington pair rated for 5 Amperes of continuous current, and with a DC current gain of 1000 minimum.
The 2n3906 is a PNP transistor (not Darlington) rated for 200 mA continuous current, and with a DC gain of 100 to 300.
For ease of understanding: They are different in multiple ways, and are designed for entirely different purposes.
That is not to say that the 2n3906 can never be used in applications somewhat similar to the one described in the Instructable: It will need a change in the schematic, and will also be limited to a far lower current than the TIP120 would be.
Let us break this down into the key differences:
- PNP v/s NPN: The TIP120 NPN device is used here as a "low side switch", in other words, when the base of the device is pulled high, the path between collector and emitter acts as a switch that conducts, thus allowing current to flow from the positive supply, via the motor or fan, through the TIP120, into ground. When the base is pulled low, the TIP120 stops conducting, the motor or fan stops operating.
To do something similar with a PNP device, one would wire it up as a "high side switch" (inverted connection), and when the base is pulled low the device would conduct (inverted logic)
- Current limits: The TIP120 will tolerate 5 Amperes of continuous current, which means a load that passes an Ampere or two at most, such as the little fan shown in the Instructable, will not cause the TIP120 any stress. Using the much less capable 2n3906 in such a task would cause it to pretty much turn into a blob of melted plastic and metal.
While it is useful at this point to also note that the two packages (TO-220 v/s TO-92) of the mentioned devices are very differently rated in terms of how much power they can safely dissipate as heat, this is actually encapsulated in the current capacity limits, so can be ignored for initial understanding.
- DC gain: The very high gain that a Darlington Pair exhibits (the gains of the two constituent BJT junctions are multiplied), allows the output to be switched quite sharply between "on" and "off" or vice versa when the base is pulled up or down past a threshold value. With the much lower gain BJT device, the switching will not be nearly as sharp as the base transitions through the threshold. The switching on or off would thus be "soggy", so to speak.
A MOSFET is another semiconductor switching device, conceptually rather different from BJTs or Darlingtons:
At the simplest level, while the BJT and Darlington show a change in current dependent on the current flowing into (or out of) the base, MOSFETs are voltage driven devices, thus the drain-to-source resistance of the MOSFET changes depending on the voltage applied to the gate.
MOSFETs can be used for switching currents to a fan or motor, just as the two devices mentioned in the question can. They just have a different set of constraints and considerations applicable - and thus a different circuit schematic to make them work.
Finally, a footnote: to-92 is a package specification for electronic components, not a specification of the actual semiconductor device contained in such a package. Often, you may find the same component available in TO-92, TO-220 and various other package specifications.