The pursuit of integrated sources for quantum optics has motivated quantum dot research for decades. However, difficulties in controlling the geometry, symmetry and size-uniformity of self-assembled quantum dots has been a major roadblock in device development. While the Stranski–Krastanov (SK) growth of InAs three-dimensional (3D) islands has been extensively investigated on GaAs(001), the SK growth mode is not observed on GaAs (110) and {111}, the other low index surfaces of GaAs. This is unfortunate as these surfaces have symmetries (Cs and C3v, respectively) which differ from that of the (001) surface, and owing to their low energy they are often present in self-assembled nanostructures such as nanowires. Furthermore, the high symmetry of {111} surfaces is expected to yield quantum dots that are ideal for entangled photon emission.Here we show that using a Bi surfactant to modify surface properties can profoundly influence epitaxial growth on GaAs (110) and (111)A surfaces. On GaAs(110), the Bi surfactant alters the fundamental growth mode of InAs from 2D layer growth to a 3D SK mode. Furthermore, a morphological phase transition can be induced “on-demand” in static strained 2D InAs(110) layers by exposing them to Bi, resulting in a rapid rearrangement of the InAs layer into 3D islands. Small (110) QDs are coherently strained to the substrate. These islands are optically active and exhibit emission that is linearly polarized, showing perspective for polarized single-photon emitters. On GaAs(111)A, GaAs buffer layer growth under a Bi flux results in ultra-smooth surfaces, which are free from the typically-observed morphological defects. Similar to the (110) case, exposing 2D InAs/GaAs(111)A layers to Bi induces the InAs 3D island self-assembly. These findings illustrate how surface-energy-modifying surfactants open the door to QDs synthesis on new substrates and with new materials.