The genesis of continental collision-related porphyry Cu deposits (PCDs) remains controversial. The most common hypothesis links their genesis with magmas derived from subduction-modified arc lithosphere. However, it is unclear whether a genetic linkage exists between collision- and subduction-related PCDs. Here, we studied Jurassic subduction-related Cu-Au and Miocene collision-related Cu-Mo porphyry deposits in south Tibet. The Jurassic PCDs occur only in the western segment of the Jurassic arc, which has depleted mantle-like isotopic compositions [e.g., (⁸⁷Sr/⁸⁶Sr)_i = 0.7041–0.7048; ε_Nd(t) as high as 7.5, and ε_Hf(t) as high as 18]. By contrast, no Jurassic PCDs have been found in the eastern arc segment, which is isotopically less juvenile [e.g., (⁸⁷Sr/⁸⁶Sr)_i = 0.7041–0.7063, ε_Nd(t) < 4.5, and ε_Hf(t) ≤ 12]. These results imply that incorporation of crustal components during underplating of Jurassic magma induced copper accumulation as sulfides at the base of the eastern Jurassic arc, inhibiting PCD formation at this time. Miocene PCDs are spatially confined to the Jurassic arc, and the giant Miocene PCDs cluster in its eastern segment where no Jurassic PCDs occur. This suggests that the arc segment barren for subduction-related PCDs could be fertile for collision-related PCDs. Miocene ore-forming porphyries have young Hf model ages and Sr-Nd-Hf isotopic compositions overlapping with those of the Jurassic rocks in the eastern segment, whereas contemporaneous barren porphyries outside the Jurassic arc have abundant zircon inheritance and crust-like Sr-Nd-Hf isotopic compositions. These data suggest that remelting of the lower crustal sulfide-bearing Cu-rich Jurassic cumulates, triggered by Cenozoic crustal thickening and/or subsequent slab break-off, led to the formation of giant Miocene PCDs. The spatial overlap and complementary metal endowment between subduction- and collision-related magmas may be used to evaluate the mineral potential for such deposits in other orogenic belts.

We have investigated a suite of Miocene high-Mg diorite porphyries from Qulong in southern Tibet, the largest porphyry Cu–Mo deposit in China. Laser ablation inductively coupled plasma mass spectrometry zircon U–Pb dating shows that the high-Mg diorite porphyry was emplaced at 15·7 ± 0·2 Ma, which is slightly younger than the Qulong adakite-like Rongmucuola pluton (c. 19·5–16·4 Ma). The Qulong high-Mg diorites have phenocrysts showing disequilibrium textures and include high-Mg# clinopyroxene (0·91–0·97). These porphyry rocks exhibit both ultrapotassic and adakite-like features, and are characterized by high contents of MgO (4·2–5·1 wt %), K₂O (3·2–3·6 wt %), and compatible trace elements (e.g. Ni: 115–142 ppm; Cr: 214–291 ppm), as well as by high Sr/Y and La/Yb ratios. The rocks have highly radiogenic isotopic compositions with (⁸⁷Sr/⁸⁶Sr)_i = 0·707004–0·707198, ε_Nd(t) = −5·1 to −5·5, ²⁰⁷Pb/²⁰⁴Pb = 15·697–15·704, and ²⁰⁸Pb/²⁰⁴Pb = 39·082–39·116, as well as variable zircon ε_Hf values of −3·0 to +5·9. Petrographic, elemental, and isotopic evidence suggests that the Qulong high-Mg diorites were formed by mixing between ultrapotassic and adakite-like melts, derived from metasomatized Tibetan lithospheric mantle and juvenile lower crust, respectively. In contrast, the Qulong pre-ore Rongmucuola pluton is characterized by high SiO₂ (66·3–68·9 wt %) and Al₂O₃ (16·4–17·0 wt %) contents, high Sr/Y ratios (121–151), low compatible element contents (e.g. Ni = 16·0–17·4 ppm; Cr = 14·5–20·2 ppm), low Mg# values (0·44–0·52), positive large-ion lithophile element (LILE) anomalies, marked negative high field strength element (HFSE) anomalies, positive ε_Nd(t) values (+0·4 to +2·5), and low (⁸⁷Sr/⁸⁶Sr)_i values (0·704847–0·705237). These features indicate that the Rongmucuola pluton was formed by partial melting of subduction-modified juvenile lower crust within the garnet stability field. The newly identified Qulong high-Mg diorite allows us to propose a mixing model for the origin of the Gangdese high-K, adakite-like rocks. In this model, the formation of these rocks occurred in two stages: (1) partial melting of highly metasomatized lithospheric mantle that generated ultrapotassic mafic melts; (2) underplating of such melts beneath thickened juvenile lower crust, which resulted in melting of the lower crust and the generation of adakite-like magmas. Mixing of the adakite-like melt with ultrapotassic magmas elevated the K₂O, MgO, and other LILE (e.g. Rb and Ba) contents of the adakite-like melt. Exogenous water necessary for formation of the Gangdese porphyry Cu systems was mainly added during mixing of ultrapotassic magma with adakite-like melt at lower- and/or upper-crustal depths, reflecting the large decrease in the H₂O solubility of the ultrapotassic mafic melt upon ascent and decompression. Upper-crustal fractionation of the Rongmucuola magma could also possibly increase the water content of mineralization-related, adakite-like porphyry intrusions at Qulong. Fluid exsolution from the ultrapotassic magma is likely to have been a key process in the generation of the Gangdese porphyry Cu deposits, as well as other porphyry Cu deposits in the Tibetan collisional orogens.