It would be more accurate to say: deepest minerals with hydroxyl groups found 1000 km down. Never at any stage will there be a fluid that can be described as

It would be more accurate to say: deepest minerals with hydroxyl groups found 1000 km down. Never at any stage will there be a fluid that can be described as free water at these depths. Breakdown of minerals with OH- such as hydrous ringwoodite at the top of the lower mantle results in dehydration melting. The melts/fluids produced will have many 10s of weight percent solutes, but in a mol. percent sense these melts/fluids are dominantly H2O, given how low the molar mass of H2O is relative to the other components. If some of this melt gets trapped inside a growing diamond, it freezes during ascent and forms minerals with structural OH-, e.g Mg(OH)2. There is ambiguity regarding what to call melts/fluids at these pressure and temperature conditions as many bulk compositions will be above the second critical end point where there is complete miscibility between H2O and 'molten rock'. Last paragraph of the paper: Our data support the idea that dehydration melting generates volatile-rich fluids/melts at the top of the lower mantle, yet being a key process for diamond genesis and for transportation of water across the 660 km discontinuity.