This paper presents the design of injection duct span width distribution of co-flow jet (CFJ) flow control airfoils in cruise condition. The duct cross section outlines are mathematically modeled as superellipse, which has a parameter η to control the outline shape. The semi major axis of the superellipse, which controls the span profile of the jet, is of great importance to the overall performance of the duct and the CFJ airfoil. The aerodynamic performance of the ducts are evaluated via numerical simulations, which employ 3D RANS solver with Spalart-Allmaras (S-A) turbulence model, 3th order WENO scheme for the inviscid fluxes, and 2nd order central differencing for the viscous terms. The duct inlet shape is predetermined according to the microcompressor performance, and the associated boundary conditions are configured based on the 2D simulation results of CFJ airfoils and micro-compressors. The design goal is to eliminate flow separation, maximize the total pressure recovery, and minimize the spanwise velocity at the injection duct outlet. The simulation results show that the duct span width distribution is crucial to control the spanwise velocity at the duct outlet. Smaller span over diameter ratio (W/D) leads to much smaller deviation angle (β) of the flow at the outlet. Moreover, a diverging-converging distribution of the duct span width leads to even smaller β angle. However, such profiles are more likely to cause flow separation at the two ends of the duct in spanwise direction. In the current study, the duct with W/D = 6.5 (design a) shows 95% total pressure recovery, while the duct with W/D = 5 (design b) shows 96.4% total pressure recovery. The first duct with diverging-converging duct span width distribution (design c) presents only half of the maximum β angle comparing to design b, but the total pressure recovery is a little bit lower (95.8%) due to flow separation at the two ends of the duct in spanwise direction. The second duct with diverging-converging duct span width distribution (design d) presents 85% of the maximum β angle comparing to design b, and the total pressure recovery is almost the same (96.3%) since the flow separation at the two ends of the design c in spanwise direction are removed. The aerodynamic performance of design d is the best in terms of low spanwise velocity and high total pressure recovery.