A semiconductor device suffering fewer current crowding effects and a method of forming the same are provided. The semiconductor device includes a substrate, a gate over the substrate, a gate spacer along an edge of the gate and overlying a portion of the substrate, a diffusion region in the substrate wherein the diffusion region comprises a first portion and a second portion between the first portion and the gate spacer. The first portion of the diffusion region has a recessed top surface. The semiconductor device further includes a silicide layer on the diffusion region, and a cap layer over at least the silicide layer. The cap layer provides a strain to the channel region of the semiconductor device.