Image sensors include a pixel die that is stacked on a logic die. The logic die includes at least one function logic element disposed on a bond side thereof, and a logic oxide array of raised logic oxide features also disposed on the bond side. The pixel die includes a pixel array disposed on a light receiving side thereof, and a pixel oxide array of raised pixel oxide features disposed on a bond side of the pixel die. A plurality of outer bonds is disposed between an outer region of the logic die and an outer region of the pixel die. A plurality of inner bonds is formed at an inner region of the image sensor between the pixel oxide array and the logic oxide array, the inner bonds being spaced apart by a plurality of fluidly connected air gaps that extend between the logic die and the pixel die.