Micro-propulsion technology is under rapid development and considerably extends the mission capabilities of small spacecraft. However, the flow in conventional nozzles on a microscale is relatively viscous, which, combined with the bounding nozzle walls, leads to low efficiencies. Aerospike nozzles, on the other hand, offer free-flow expansion and mitigate the boundary layer effects. The research objective of this project is to investigate the losses and output efficiency of micro-nozzles by optimizing a 3D axisymmetric aerospike in the μN-mN range. Planar nozzles, typically found in micro- propulsion subsystems, limit the flow radial expansion and considerably suffer from viscous losses, as well as momentum loss. In addition, in lower thrust settings, the boundary layer massively limits the performance of conventional nozzles – conical and bell layouts –, whereas aerospikes do not bound the flow and, subsequently, inherit better gas expansion. The design optimization relies on numerical analyses, with the free and open-source OpenFOAM’s solver, rhoCentralFoam, and self-developed Python scripts that automate the simulations in various machines. This study comprises three design iterations. The first contour parameters derive from Delft University of Technology’s Vaporizing Liquid Micro-thruster (VLM) requirements and the preceding work. However, since this work considers axisymmetric geometries, the aerospike’s throat width is reduced from 45 μm to 15 μm to preserve the original thrust magnitude (< 10 mN). The initial results show that, at the same throat Reynolds number, the aerospike outperforms the bell nozzle, especially towards lower Reynolds, where the specific impulse efficiency variation tops 25%. However, at equal thrust levels, the conventional design surpasses the aerospike thrust and specific impulse efficiencies. The Mach contours reveal that the small throat width and high area ratio ineffectively mitigate the viscous losses and lead to extreme momentum loss. The following iterations with four truncation percentages (20%, 40%, 60%, and 80%) prove the first hypothesis right: when decreasing the area ratio from ~17 to ~3 and raising the throat width to 30 μm, the efficiency of an 80% long aerospike reaches ~94% for the specific impulse and ~89% for the thrust. In addition, the aerospike yields the best performance when it is the least truncated (highest truncation percentage, i.e., 80%). Finally, a ±100 K temperature sensitivity study shows that the aerospike performance oscillates up to 3%, with a maximum thrust efficiency of 91% and specific impulse efficiency close to 98%, rivaling macroscale performance. With a small area ratio and a wide throat, the aerospike nozzle outperforms an equivalent bell nozzle by more than 10% in terms of specific impulse and thrust efficiencies.