A stacked semiconductor device includes a CMOS device and a MEMS device. The CMOS device includes a multilayer interconnect with metal elements disposed over the multilayer interconnect. The MEMS device includes metal sections with a first dielectric layer disposed over the metal sections. A cavity in the first dielectric layer exposes portions of the metal sections. A dielectric stop layer is disposed at least over the interior surface of the cavity. A movable structure is disposed over a front surface of the first dielectric layer and suspending over the cavity. The movable structure includes a second dielectric layer over the front surface of the first dielectric layer and suspending over the cavity, metal features over the second dielectric layer, and a flexible dielectric membrane over the metal features. The CMOS device is bonded to the MEMS device with the metal elements toward the flexible dielectric membrane.