For a 0.5µm process with double poly option you have a specific capacitance of 0.9fF/(µm)², so you'd need an area A = 22000/0.9 = 24444(µm)² or a square cap of 156*156(µm)² - which is huge!
You could use a MOSCAP, which - for a 0.5µm process with tox=13.5nm - has a specific capacitance of 2.56fF/(µm)². With that, your 22pF cap would need an area A = 22000/2.56 = 8594(µm)² or a square cap of 93*93(µm)² - which is still huge, but perhaps feasible.
A MOSCAP has a voltage-dependent capacity, and for the a.m. specific capacitance it needs a voltage difference of > |Vth|.
It's probably worth to think twice if you really need such a big capacitance. Perhaps you can decrease it by more design ideation?