Effect of water on carbonate-silicate liquid immiscibility in the system KAlSi (sub 3) O (sub 8) -CaMgSi (sub 2) O (sub 6) -NaAlSi (sub 2) O (sub 6) -CaMg(CO (sub 3) ) (sub 2) at 6 GPa; implications for diamond-forming melts
Effect of water on carbonate-silicate liquid immiscibility in the system KAlSi (sub 3) O (sub 8) -CaMgSi (sub 2) O (sub 6) -NaAlSi (sub 2) O (sub 6) -CaMg(CO (sub 3) ) (sub 2) at 6 GPa; implications for diamond-forming melts
American Mineralogist (February 2021) 106 (2): 165-173
- alkali feldspar
- carbonates
- carbonatites
- chain silicates
- clinopyroxene
- coesite
- diamond
- diopside
- dolomite
- experimental studies
- feldspar group
- fluid inclusions
- fluid phase
- framework silicates
- garnet group
- high pressure
- igneous rocks
- immiscibility
- inclusions
- K-feldspar
- kyanite
- liquid phase
- mantle
- melt inclusions
- melts
- mica group
- native elements
- nesosilicates
- orthosilicates
- phase equilibria
- phengite
- pressure
- pyroxene group
- sheet silicates
- silica minerals
- silicate melts
- silicates
- upper mantle
To clarify the effect of water on carbonate-silicate liquid immiscibility in the diamond stability field, we performed experiments in the system KAlSi (sub 3) O (sub 8) -CaMgSi (sub 2) O (sub 6) -NaAlSi (sub 2) O (sub 6) -CaMg(CO (sub 3) ) (sub 2) under nominally dry and hydrous conditions by adding 1.5 wt% H (sub 2) O at a pressure of 6 GPa and temperatures of 1000 to 1500 degrees C. Both systems start to melt at 1050-1100 degrees C. Under anhydrous condition the melting occurs via the following reaction: 6KAlSi (sub 3) O (sub 8) (K-feldspar) + 6CaMg(CO (sub 3) ) (sub 2) (dolomite)=2(Ca (sub n) Mg (sub 1-n) ) (sub 3) Al (sub 2) Si (sub 3) O (sub 12) (garnet) + Al (sub 2) SiO (sub 5) (kyanite) + 11SiO (sub 2) (coesite) + 3 K (sub 2) (Ca (sub 1-n) , Mg (sub n) ) (sub 2) (CO (sub 3) ) (sub 3) (carbonatitic melt) + 3CO (sub 2) (fluid and/or liquid), where n approximately 0.3-0.4. The carbonatitic melt has the following composition 38(K (sub 0.92) Na (sub 0.08) ) (sub 2) CO (sub 3) .62Ca (sub 0.62) Mg (sub 0.38) CO (sub 3) . A second immiscible silicic melt containing (in wt%, volatile free) SiO (sub 2) =68.8, Al (sub 2) O (sub 3) =12.6, CaO=3.7, MgO=2.4, Na (sub 2) O=1.1, and K (sub 2) O=11.3 appears at 1250 degrees C. Both melts remain stable up to 1500 degrees C and coexist with the clinopyroxene+ or -garnet+ or -coesite residue. In the presence of water stored away in phengite, the melting begins with silicic melt, which contains (in wt%, volatile free) SiO (sub 2) =61.4, Al (sub 2) O (sub 3) =15.3, CaO=4.8, MgO=3.0, Na (sub 2) O=2.2, and K (sub 2) O=13.3, and coexists with phengite, dolomite, clinopyroxene, and coesite. The phengite + dolomite assemblage remains to 1100 degrees C and disappears at 1200 degrees C producing two immiscible melts carbonatitic with approximate composition, 19(K (sub 0.89) Na (sub 0.11) ) (sub 2) CO (sub 3) .81Ca (sub 0.57) Mg (sub 0.43) CO (sub 3) , and silicic containing (in wt%, volatile free) SiO (sub 2) =63.3, Al (sub 2) O (sub 3) =15.6, CaO=4.5, MgO=3.0, Na (sub 2) O=2.0, K (sub 2) O=11.6. The present results imply that partial melting of continental material subducted to a depth of 200 km can yield simultaneous formation of two immiscible melts, K-dolomitic and K-aluminosilicate. Under dry conditions, carbonatitic melt appears earlier (at a lower temperature). Given the low density and high mobility of this melt, it must rapidly percolate upward, leaving a refractory eclogite-like residue and leaving no chance for the formation of a second aluminosilicate melt. However, under hydrous conditions silicate melt appears earlier than carbonatitic melt, leaving a phengite- and dolomite-bearing residue, which finally yields the formation of two immiscible silicic and carbonatitic melts. The compositions of these melts fall in the compositional range of carbonatitic and silicic high-density fluids (HDFs) in diamonds worldwide. Thus, we suggest that the presence of water is a necessary requirement for the formation of immiscible HDFs inclusions in diamonds, and this suggestion is strongly supported by natural data from HDFs.