The Changbaishan volcano is well known for its major caldera-forming Millennium Eruption (ME) in 946 CE (Common Era). We report Hf–O isotopes of zircon grains from pre-caldera Qixiangzhan (QXZ) and syn-caldera eruptions of the Changbaishan (Baitoushan) volcano to constrain magma chamber processes. Zircon grains from the pre-caldera QXZ comendite lavas have δ18O ranging from 4.46 to 5.16 (lower than mantle values) and εHf ranging from −4.47 to +4.37. Zircon grains from the syn-caldera ME1 charcoal-bearing non-welded comendite pyroclastic flow deposits have δ18O ranging from 2.25 (lower than mantle values) to 5.51 and εHf from −3.75 to +3.31. By comparison, zircon grains from the ME2 welded trachytes have δ18O ranging from 5.66 to 6.20 (higher than mantle zircon values) and εHf from −1.97 to +6.23. There are no correlations between O and Hf isotopes for all zircon grains in QXZ and ME1 comendites and ME2 trachyte. The ubiquitous occurrence of low-δ18O zircon grains in QXZ and ME1 comendites indicates shallow remelting of hydrothermally altered low-δ18O juvenile rocks. By contrast, ME2 trachyte zircons (except for two zircon grains) have normal δ18O (5.66 to 6.10) values, indicating a lack of remelting processes. Similar zircon Hf–O isotopes between pre-caldera QXZ comendites and syn-caldera ME1 comendites indicate tapping of the upper portion of a zoned magma chamber. Higher δ18O in ME2 trachyte zircons indicate tapping of the deeper portion of a zoned magma chamber free from shallow remelting. The lack of significant correlations between zircon O and Hf isotopes, and the relatively high εHf values for all Changbai zircon grains, argue against partial melting of ancient continental crust or significant contaminations by ancient crustal rocks as an origin for these felsic magmas. The QXZ and ME1 comendites were formed by shallow remelting of hydrothermally altered juvenile volcanic rocks, and ME2 trachytes were formed by evolution of mantle-derived basaltic magmas free of hydrothermal assimilations. A proto-caldera likely formed prior to the generation of QXZ lavas at 10 ka.
Read full abstract