Difference between a constant DC and PWM

What is the purpose of R2 in the first circuit? It's limiting the maximum current to the motor to 12V/220Ω = 54.5mA, which is probably not enough.

Just connect the motor directly to the collector of Q2.

You also need a couple of other changes: put a resistor (e.g., 4700Ω) between the MCU output and the base of Q1; otherwise, the B-E junction of Q1 looks pretty uch like a dead short to ground to the MCU. Similarly, you need a resistor (about 1200Ω) between the collector of Q1 and the base of Q2; without it, when Q1 switches on, it basically shorts the 12V supply to ground via the B-E junction of Q2. 1200Ω limits the base drive to Q2 to about 10mA, which should be plenty.

First of all I would strongly recommend you to add a parallel diode to the motor with its anode connected to MOSFET's drain and with its cathode to +12V. This diode should be schotky (1N5819) or fast recovery silicon (FRxxx, HERxxx, SFxx). This is done to prevent the inductive voltage overshoot of motor's winding right after MOSFET is turned off.

Second - your parallel capacitor 100nF is too big for PWM control. Each cycle the transistor is turned on a large amount of current will pass through this cap and heat up the transistor. A value of 1nF to 10nF maximum would be OK. If there is plenty emitted noise due to motor brushes (this is the main function of this cap) you could place a resistor in series with the capacitor - something like a 10-47 ohm. This way the stress on the MOSFET will be reduced significantly while the capacitor still does its job with motor brushes.

Measuring 8.7V at the drain makes me think that there are significant voltage overshoot after each turn off and thus the average value is high. Measure that again after placing the diode.