Because the Base-Emitter of an NPN Bipolar transistor looks like a forward biased diode (current-operated), where substantial current starts to flow once the voltage reaches ~0.65V, timers based on a RC network do not work well, or the delay must be short ,i.e. during the time the capacitor charges from an initial voltage, typically 0V, to 0.65V.
Using the bipolar transistor puts a lower bound on R; the current through R must be sufficient to turn on the transistor. In practical circuits, R cannot be more than ~100KΩ.
If you replace the NPN with an NFET (voltage operated), then you get much longer, accurate delays. This is because no significant current flows into the Gate-Source of the NFET; all of the current through the timing resistor R goes into charging the capacitor C. The turn-on threshold of an enhancement-mode FET is usually several volts, making for a longer delay period.
Using the FET allows R to be up to tens of megΩ. The upper limit comes from capacitor leakage current.
Try building a timer with a delay of 10 minutes...