Copper is one of the most conductive metals known to humankind, making it the cornerstone for all electricity related technologies (e.g., cars, power generation, household appliances). This is causing an ever-increasing global demand for copper to support society's efforts for a low-carbon future. 

Most of the world's copper is derived from porphyry copper deposits (PCDs), where magmatic-hydrothermal fluids cause copper-ore mineralization centered on a granitic intrusion. The prevailing view of the formation of porphyry copper deposits along convergent plate boundaries involves deep crustal differentiation of metal-bearing juvenile magmas derived from the mantle wedge above a subduction zone. However, many major porphyry copper districts such as La Escondida and Rio Blanco-Los Bronces in Chile, and Resolution in Arizona, formed during periods of flat-slab subduction when the mantle wedge would have been reduced or absent, leaving the source of the ore-forming magmas unclear (Figure 1).

Upon starting my BHP funded postdoc at the University of Bristol in 2021, under the supervision of Frances Cooper, our original objective was to constrain the geodynamic controls of porphyry copper mineralization in the Laramide Porphyry Province of Arizona, SW USA. Being completely new to working on porphyry copper deposits and the geology of the SW USA, I approached the project thinking like a metamorphic/ igneous petrologist and trying to get my head around the large scale tectonic processes associated with the Laramide Orogeny.

The first thing that struck me when studying the map of Arizona was the contemporaneous contrasting granite types - Biotite±Hornblende bearing granites and more peraluminous Muscovite±Garnet bearing granites, with the later representing crustal derived magmas based on their isotopic signatures. Interestingly many of the Biotite±Hornblende bearing granites that were associated with most  PCDs also show similar isotopic signatures from the literature, implying a significant crustal contribution in the source(s) to these magmas. Having researched the granite magmatism and crustal melting in the Aegean during my PhD and previous postdoc, I was intrigued by this observation and hypothesized whether the contrasting granite types could have formed by a similar process, i.e., melting mafic and more felsic/pelitic crustal protoliths respectively.

Upon Covid restrictions relaxing in November 2021, we (Frances and myself) were able to undertake fieldwork in Arizona, and we met up with collaborators from the Natural History Museum, BHP and University of Western Australia who were researching different aspects of porphyry copper mineralization processes. We undertook a week long workshop and fieldtrip visiting different copper deposits in SE Arizona, and several Laramide thrust faults. One of the key observations we wanted to explore was the relationship between compressional tectonism, ending of compression, volcanism and mineralization.

After a few days making observations, Matthew Loader (NHM), Jamie Wilkinson (NHM) and myself noted the potential importance of crustal melting in forming the Laramide granites, and that both Biotite±Hornblende bearing granites and  Muscovite±Garnet bearing granites were magnetite bearing, had elevated Sr/Y 'adakite like' geochemistry  and an abundance of Proterozoic zircon inheritance. Matt and myself then undertook 4 weeks of fieldwork sampling the Laramide intrusions and their relationship with thrust faults from SE Arizona near the Mexican border all the way up to Las Vegas in Nevada - an epic roadtrip and once in a lifetime adventure (Figure 2). During this fieldtrip, it became clear that porphyry copper formation immediately post-dated the local relaxation of compression and volcanism, but the timing of compressional tectonism, volcanism and mineralization seemed to get younger from NW Arizona at ~ 73-70 Ma to SE Arizona ~ 60-56 Ma. Furthermore, both granite types accompanied porphyry copper mineralization to various degrees!

We also visited garnet-clinoproxene xenoliths at Camp Creek and Chino Valley that had and abundance of visible sulphides, and likely represent the Arizona lower crust, as well as the Harcuvar, Harquahala and Granite Wash Mountains metamorphic core complexes (exposures of the mid-lower crust exhumed beneath a low angle normal fault or ductile shear zone during the Basin and range extension). I was pleasantly surprised to find migmatites and high grade metamorphic rocks. However, the age and petrogenesis of these rocks were poorly known although the Tank Pass muscovite-biotite-garnet bearing granite (ca. 78-56 Ma) seemed to be co-genetic with the partial melting and copper mineralization even within the metamorphic rocks (Figure 3). 

Upon returning to Bristol and intrigued by the potential importance of crustal melting in forming the porphyry copper deposits Arizona, I compiled all available literature isotopic, geochronological, geochemical and tectonic data from the SW USA. The results demonstrated that the Laramide intrusions overlapped with the known Proterozoic basement from the area (Figure 4). Furthermore, we (Jamie, Matthew and myself) realised that there was copper mineralization in Arizona dating back to the Proterozoic with the 1.7 Ga Squaw Peak deposit and Jerome volcanogenic massive sulphide deposit, which could be important for the regional metallurgy. Integrating these results with the recent discovery of a subduction melange representing the Farallon slab at Cemetery Ridge in Arizona that yielded metamorphic zircon ages of ca. 70-40 Ma, together with the occurrence of Lawsonite bearing eclogite xenoliths at the centre of the Colorado Plateau indicated that flat-slab subduction occurred beneath Arizona at the time of porphyry copper mineralization. I went back to Arizona for another two field seasons to specifically investigate this question and to constrain cross-cutting relations and collect samples of the migmatites, thrust faults and granites. 

Upon analysing the migmatitic samples from the Harcuvar mountains, another key observation became apparent. The melt composition of the migmatites was trondjehmitic, with very little peritectic K-feldspar, yet extensive amounts of leucosome (former melt ~>35%) despite the igneous nature of the basement protolith. However this would be consistent with water-fluxed melting. The addition of water facilitates the breakdown of plagioclase and increases the stability of amphibole as a restitic phase - this could also explain the high Sr/Y ratios that characterize the Laramide granites (Sr is consumed by plagioclase - so more plagioclase melting increases the magma Sr/Y)!

We then dated the migmatites via U-Th-Pb monazite geochronology (Nick Roberts at the British Geological Survey) and in-situ Rb-Sr  geochronology (Dan Bevan, Tim Elliot). We found the timing of partial melting (ca. 73-60 Ma) overlapped precisely with the timing of granitic magmatism, metamorphism of the Farallon plate and porphyry copper mineralization. This was followed by cooling of rock between 60-20 Ma (Figure 5)! It all started to make sense!

What I hypothesized two years earlier was becoming consistent with these data. Water was fluxing off the flat-slab, but instead of melting the mantle wedge which was removed by the flat-slab, water advects across the Moho. This reduced the solidus and caused melting of different Proterozoic crustal rocks and forming migmatites and contrasting types of granites with a 'crustal' isotopic signature.

However, as porphyry copper mineralization was a transient feature, occurring a few Myrs at any given location after arc-volcanism shuts down and compression ends, and was followed by tectonic and magmatic quiescence yet gets younger from NW to SE Arizona, it is likely related to an underlying geodynamic process. After some thought and conversations with Adam Gorecki (BHP), it became clear that onset of flat-slab subduction related to oblique collision of the Hess oceanic plateau conjugate could explain this observation (Figure 6). This is because the onset of slab-flattening would shut-down the mantle wedge, and allow water to flux into hot lower crust, causing water fluxed crustal melting and may have potentially reduced the crustal strength causing a localised change in stress regime. However continued underthrusting of the flat-slab would have caused cooling of the overriding North American lithosphere, explaining the post mineralization tectonic and magmatic quiescence. The oblique Hess plateau collision would therefore cause diachronous slab flattening from NW-SE Arizona between 73-56 Ma. Interestingly a similar phenomenon occurs in Chile associated with the Miocene to present day subduction of the Juan de Fernadez Ridge, which was associated with sequential shut-down of the volcanic arc followed by porphyry copper formation that get younger towards the south.

We therefore propose that flat-slab subduction is fundamental to fluxing water into the base of the lower crust causing melting and remobilizing metals that accumulated during previous periods of arc magmatism. In conclusion, we suggest that flat-slab regimes should be targeted for future mineral exploration to help support society's ever increasing demand for copper.

Find our paper on the link: https://www.nature.com/articles/s41561-024-01575-2