Getting 3V sounds weird to me. I'd measure voltages in the circuit. Like, gate voltage with respect to ground, voltage across mosfet, and voltage between gate and source.
Also, I'm not sure ripple should completely disappear. I think it only attenuates it. I didn't do measurements on mosfet cap multipliers, but with with 2n3904 I only got some modest attenuation, like by a factor of 20-40x. Probably, darlington will work better at modest frequencies.
I'd also prefer darlington because, imho, it should give lower voltage dropout. With mosfet the dropout voltage will be more than Vgs(th), which is typically 2-3V. Darlington would give 1.3-1.5V dropout (my estimation).
PS do you have an oscilloscope?