A mechanism for ocean-floor spreading

I wish to suggest a mechanism for ocean-floor spreading, other than deep mantle convection cells. If the geotherm intersects the mantle solidus at the base of the upper mantle, a partly melted layer will result. It will be less dense than the overlying unmelted garnet peridotite lithosphere and the situation will be gravitationally unstable. It will be more voluminous than the overlying unmelted mantle and will distend the lithosphere which will break up into plates. It will have low rigidity and decouple those lithospheric plates from the underlying transition zone and lower mantle. Melted material, perhaps with crystals, will escape in two ways.