The existence of subterranean oceans is not based in reality at all. All this reference to oceans in the mantle is science journalist bullshit. There is no ocea

The existence of subterranean oceans is not based in reality at all. All this reference to oceans in the mantle is science journalist bullshit. There is no ocean at 700 km depth. The paper refers to the mineral ringwoodite, for which the anhydrous formula is (Fe,Mg)SiO4. Hydrous ringwoodite has the capacity to take some OH groups in the mineral structure, probably in a coupled substitution with vacancies/defects/some trace element such as titanium. Recently a ringwoodite sample was found that had 1.5 weight % H2O in its structure. What that means is it is entirely unlikely that there will ever be free water in the mantle, because you can dissolve so much of it in the minerals. Free water can only exist when the minerals can not longer fit any more in them. What the authors did was take some hydrous ringwoodite and put it in a diamond anvil cell and heat a bit of it with a laser. The bit they heated transformed into the minerals perovskite and ferropericlase, and amorphous material the authors interpret as quenched silicate liquid (i.e. glass, analogous to obsidian). The perovskite and ferropericlase have less capacity to dissolve H2O than the ringwoodite, so when ringwoodite breaks down, some H2O is releasd. However, instead of forming free water, the H2O reacts with the other minerals to form a silicate liquid (i.e. magma with some H2O dissolved in it). This reaction is referred to as dehydration melting: the reaction of a hydrous mineral phase to form other mineral phases and a silicate liquid. You dehydrate a mineral and form melt. The authors then relate the mineral scale reactions to global scale seismic structure. Low velocity regions are observed when mantle downwells from above the 660 km continuity to below it. Low velocity zones are interpreted to be associated with the presence of silicate liquids. Thus, the authors suggest that downwelling of hydrous ringwoodite bearing mantle from above the 660 discontinuity to below it will result in a phase change to pervoskite + ferropericlase + silicate liquid, consistent with the presence of low velocity zones. The amount of silicate liquid produced is about 1%. You don't need much silicate liquid to make low velocity zones. This entire story does not involve free H2O at any stage. Any reference to oceans is misleading.