That’s quick. On a 0.2ms timescale, a 22ns rise and fall time is not very noticeable. I have the PWM pin connected to a 200 ohms resistor, and then trough a 4N35 opto.
Actually I measure something closer to 2ns rise/fall time (on a 20 GHz sampling scope). That is when going to an ohmic load. When you drive a long line and/or connect a (capacitive) input of another chip to it the rise time will be slower.