Previous work has shown that under quasi-isentropic compression above 2.5 GPa water may enter the phase region of ice VII. The liquid remains mostly metastable whilst ice nucleates sporadically over ~100 ns. The phase change occurred reliably when water was compressed between silica surfaces but was not observed when these were made from sapphire. Through further quasi-isentropic compression experiments on a gas gun this work has determined that the nucleation initiates at the window-water interface and that initiation likely depends on the surface energy of the silica. Sapphire, aluminium and reactively sputtered silica did not nucleate the phase change and the water remained metastable. The limit of this metastable liquid water was thought to be at 7 GPa. Through experiments with both silica and sapphire the threshold was found to be closer to 6.5 GPa and appears to cause a homogeneous phase change nucleating throughout the volume of water and independent of surfaces.