Dynamic feldspar alteration governed by single self-evolved uid system

The evolution of hydrothermal uids during metasomatic and/or hydrothermal processes is responsible for the formation of ore deposits and associated alteration. In systems with well-developed breccia and fractures, mineral reactions are largely driven by decompression boiling, uid cooling or external uid mixing, but in less permeable rocks, elements exchanges occur at uid-mineral interfaces, resulting in a self-evolved uid-mineral reaction system. However, the dynamic uid evolution leading to large-scale (km) alteration remains poorly understood. We observed experimentally that the sequential sodic and potassic alterations associated with mineralization in large ore deposits, in particular Iron Oxide Copper Gold (IOCG) deposits, can occur via a single self-evolved, originally Na-only, hydrothermal uid, driven by a positive feedback between equilibrium and kinetic factors. Albite formed rst upon reaction of sanidine ((K,Na)AlSi3O8) with a NaCl uid at 600˚C, 2 kbar. However, with increasing reaction time, some of the initially formed albite was in-turn replaced by K-feldspar (KAlSi3O8). Fluorine accelerated the process, resulting in nearly complete back-replacement of albite within 1 day. These experiments demonstrate that potassic alteration can be induced by Na-rich uids, and pervasive sequential sodic and potassic alterations do not necessarily reect near-equilibrium, externally-driven changes in uid alkali contents.


Introduction
Fluid-rock interactions during metasomatic and/or hydrothermal processes control crustal rheology, porosity structure, and elements redistribution within the Earth's crust 1 . In mineral systems, the ore itself represents of tiny volume within an extensive (10's to 100's of km 3 ) hydrothermal system, and many large deposits, in particular IOCG 2 and porphyry 3 , are associated with extensive (tens to hundreds of km 3 ) alteration halos [4][5][6] . These halos form as a result of thermodynamic disequilibrium between rock and uid, and many of the characteristic mineral reactions are driven by alkali (Na,K) exchange. An evolution from sodic to potassic alteration is a key feature of IOCG deposits. The current consensus is that this evolution is caused by changing chemical and/or physical conditions 2,7 (e.g., cooling; decompression boiling; decreasing water to rock ratio), and can be well approximated as a (near-) equilibrium system 8 .
However, recent progress in our understanding of the mechanisms of uid-driven mineral reactions has highlighted the signi cance of kinetic factors and local equilibria in controlling the evolution of uidmineral systems 9 . Interfacial uids at uid-mineral boundaries control reaction kinetics and mineral stability, and the complex feedback between uid ow, reaction progress, and transient, reaction-induced porosity is a key driver of crustal-scale uid-rock interaction 10-12 . We hypothesize that the widespread temporal and spatial association between sodic and potassic alteration may be facilitated by an interplay of kinetic and equilibrium thermodynamic factors at the uidmineral interface. To test this idea, we examined the alteration of a mixed Na-K-feldspar (sanidine) in a self-evolved, originally Na-only hydrothermal uid (NaCl or NaF) in closed system experiments at isothermal, isobaric conditions. Since feldspars make up more than 50% of Earth's crust, their complex compositions and/or textures can shed light on the crustal thermal and alteration history 13,14 . Pioneering experiments investigated the microstructural and chemical evolution of feldspars under hydrothermal conditions [15][16][17][18] , and concluded that the reactions proceeded via a uid-driven interface-coupled dissolution-reprecipitation mechanism, but the reactions products were broadly in-line with predictions from equilibrium thermodynamics. In contrast, we observed a remarkable sequential sodic and potassic alteration of sanidine rather than the expected albitisation. The replacement kinetics also increased dramatically in F-rich systems. Based on the characterization of the products and thermodynamic modelling of the uid-mineral interaction, we show that sequential sodic and potassic alterations can be driven by Na-rich uids at regional scale (i.e., not much K in the uid is required to stabilise K-feldspar versus albite), and promoted by a self-evolution process at the micro-scale.

Reaction Textures And Products
The main reaction products from both NaCl and NaF solutions are albite and/or K-feldspar ( Fig. 1; Table   S2); small amounts of biotite formed in NaCl-only solutions, and uorite and ilmenite in NaF-bearing solutions (Fig. S1). Fluorite and biotite occur mainly along the reaction front between albite and sanidine or lling pores.
In NaCl-solutions, the overall reactions proceeded through three stages. Stage I: an albite rim replaces the outmost part of sanidine ( Fig. 2A). There can be large (5-20 µm wide) gaps between the reaction rim and the pristine sanidine or a sharp interface between them ( Fig. 2A). Stage II: a new K-feldspar appears (Fig. 2B), replacing the albite formed in stage I from the outside of the grain. The interface between the new K-feldspar and albite is sharp without any micro-scale porosity. Stage III: as the reaction proceeds from rim to core, thin K-feldspar rim forms via partial replacement of the albite rim (Fig. 2C). The new Kfeldspar is characterized by higher K and lower Na contents than the starting sanidine (Figs. 2GH, S2).
However, only small amounts of the new K-feldspar phase form, and this quantity increases only slightly over time (0.08 vol% after 3 days, and 0.41 vol% after 5 days; Fig. 1A; Table S2). The reaction process was greatly accelerated in the F-bearing system ( grains nucleating on the surface of the sanidine seeds or growing into the solution were only observed in F-bearing solutions (Fig. 2F). In a few areas, K-feldspar directly replaces sanidine ( Fig. 2F insert), in which case the albite rim is fully back-reacted. Hence, in these experiments, potassic alteration is induced without any external K-input, via dynamic evolution driven by an initial disequilibrium between mineral and Na-only uid.

Thermodynamic Modelling
The aim of this section is to predict the system evolution under equilibrium conditions, and then use these predictions to identify potential kinetic effects that may explain the observed sequential sodic and potassic alterations. Thermodynamic calculations are challenging since our experiments involve reaction between a complex electrolyte solution with evolving composition and mineral solid-solutions with a miscibility gap (Fig. 3A). We used GEM-Selektor (details in the Supplementary Data), since the suitability of this package to effectively model this complex thermodynamic situation has been demonstrated 23,24 . At 600 ˚C, the sanidine composition used in our experiments is within the miscibility gap between Kfeldspar and albite, though it is close to the single-phase boundary (Fig. 3A). The Lippmann diagram in  . 3D), dissolution of sanidine quickly results in equilibration of the solution with a feldspar of composition close to that of the titrated sanidine. In the NaCl solution (Fig. 3E), albite is predicted to form rst, resulting in decreased Na + and increased K + in solution. When the sanidine/water molar ratio is around 0.25, two different products (Na-poor K-feldspar and a Na-rich albite) formed at the peritectic point. The NaF solution showed a similar evolution, but the equilibrium peritectic point was reached at lower sanidine/water ratio (Fig. 3F).
Hence, the equilibrium simulations show that unless sanidine dissolves congruently and buffers uid composition (Fig. 3D), sanidine should be replaced by two feldspars at the peritectic point (Figs. 3E-F), with a nal uid composition with X(Na + ,aq) = 0.85. These predictions tally with previous experiments 18 , whereby an Ab 60 Or 40 feldspar was replaced by coarse-grained, coexisting albite and K-feldspar upon reaction with H 2 O/HCl solutions (Fig. 3G).
However, our experiments display different textures and nal product compositions than those predicted by equilibrium thermodynamic modelling (Figs. 2, 3H). The most signi cant difference is that there is no evidence for co-precipitation of two feldspars in our experiments: albite rst replaces sanidine, and then this newly formed albite -and some sanidine-are replaced by K-rich feldspar along a separate, independent reaction front (Fig. 3H). This indicates that mineral formation in this system is governed by interfacial uids at the reaction fronts with compositions different from the bulk solution.

Reaction Mechanism
The formation of biotite, uorite and ilmenite coexisting with albite tallies with the complete dissolution of sanidine releasing the minor amounts of incorporated Ca, Ti, and Fe. The sharp boundaries between sanidine and albite/K-feldspar and the pseudomorphic replacement are characteristic of the ICDR reaction mechanism 9 . This ICDR mechanism was further con rmed by experiments conducted using isotopically tagged ( (Figs. 4, Fig. S4). This suggests that K-feldspar undergoes continuous dynamic recrystallization, which is recorded by uptake of increasing amounts of 16 O from dissolved sanidine over time. Such a dynamic recrystallization implies that the original composition (elemental and isotopic) and texture (including porosity) of the newly formed feldspars can evolve rapidly (hours) in contact with a uid.
Fluorine increased the rate of back-reaction, but the reaction mechanisms were similar regardless of Cl and F availability: sanidine dissolution, initial albitization, followed by back-replacement by K-feldspar.
Initial sanidine dissolution results in the formation of a solution surface layer with elevated K + /Na + ratio relative to the bulk solution; however, albite formed rst, in accordance with the initial bulk uid composition.
As the reaction rim expands, chemical exchanges between the reaction front and the bulk solution occur either through (transient) reaction-induced porosity 10 , or in the absence of a connected porosity network as in our experiments, along the reaction interface 15,16 . As the albite rim becomes thicker, the removal of K + from the interface and the supply of Na + from bulk uid to form albite are expected to slow down, resulting in the local enrichment of K + . However, albite with little change in composition continues to precipitate, and K-feldspar never nucleates at the reaction front between albite and sanidine. Instead, Kfeldspar starts replacing albite from the outside of the grains, forming a second, decoupled reaction interface (Fig. 2) in contrast to the co-precipitation of albite and K-feldspar from thermodynamic modelling. The newly formed K-feldspar is inhomogeneous and characterised by a lower Na/(Na + K) ratio (0.13 ~ 0.31 with average 0.22) than pristine sanidine (~ 0.35) (Fig. S2). This wide composition range of K-feldspar implies a continuously evolving interfacial uid which may become variably enriched in K + throughout the reaction. Altogether, these observations suggest a strong link between nucleation and feldspar composition: once albite nucleates with a particular composition, it continues to grow, irrespective of interface uid composition. Then a similar process happens with nucleation of a K-rich feldspar, leading to near complete replacement (back reaction) of the earlier formed albite. Hence, the back-reaction responsible of K-feldspar formation is the result of kinetic processes that prevent the equilibrium co-precipitation of two feldspars under the conditions of our experiments.

Pervasive Successive Feldspar Replacement In Nature
Many ore-forming systems, including some porphyry 4,19 and most IOCG deposits 28 are associated with large-scale sodic alteration overprinted by potassic alteration 31 . Understanding the dynamic evolution of hydrothermal uids in these alteration halos is the focus of a large body of research aiming at de ning the sources of precious and base metals, as well as identifying geochemical and mineralogical indicators for guiding mineral exploration [26][27] . In general, successive alteration styles are explained as re ecting either (i) changing uid compositions over time, (ii) temperature changes in externally-derived convective systems 26,29 ; (iii) Na-addition via interaction with evaporite-bearing wall rocks 30 ; or (iv) magmatic unmixing of H 2 O-CO 2 -NaCl ± CaCl 2 -KCl uids caused by decreases in temperature and/or pressure 31 .
However, our experimental and thermodynamic results show that this alteration sequence can also be achieved through a single self-evolved NaCl-rich hydrothermal uid at constant temperature and pressure.
In the K-feldspar alteration envelope of the Butte porphyry system, X(Na + ,aq) ranges from 0.77 to 0.85 for hydrothermal uid at 400-840 °C 20 . These uid compositions are typical for magmatic uids, that are originally rich in NaCl 32 . Thermodynamic modelling indicates that these uids lie near the composition of a uid equilibrated with albite and K-feldspar (X(Na + ,aq) ~ 0.85 @ 600˚C; Fig. 3), and could initiate proximal albitisation or potassic alteration, depending of local conditions. Albitisation reactions result in a decreasing X(Na + ,aq), and a small decrease would result in the formation of K-feldspar. However, equilibrium thermodynamic modelling suggests that potassic alteration in such a steady state system would be limited, and would feature co-existence of albite and K-felspar, contrary to eld observations of early albite and late potassic alteration. Our experiments reveal that kinetic factors acting at the reaction interface may be important in forming potassium alteration zones surrounding and/or above the sodic alteration zone in such systems. The energetic barriers that need to be overcome to change from Kfeldspar nucleation and growth to albite nucleation effectively result in the formation of K-feldspar even past the point where the uid at the reaction front becomes supersaturated with respect to albite due of increasing Na/K ratios caused by K-feldspar precipitation. This maintains a high level of disequilibrium at the reaction front, which in turns contributes to pervasive mineral replacement.
In conclusion, sequential sodic and potassic alterations may be controlled by a self-evolved originally Naenriched hydrothermal system, instead of externally driven chemical or physical changes. In nature, this internally-driven process most likely works in tandem with external drivers (e.g., the addition of external uids, decrease of temperature and/or pressure, uid unmixing). Plümper et al. 10 recently highlighted the importance of the coupling between nano-scale reaction mechanism, that forms a transient reactioninduced porosity, and macro-scale uid ow for explaining pervasive crustal, uid-present reactions (e.g., metamorphism; dolomitisation). Our results show that another type of coupling between an internally driven kinetic process at the nano-to micro-scale, and externally driven factors (predictable using equilibrium thermodynamics) at the macro-scale, may contribute to explaining the widespread occurrence of albitisation as well as the common occurrence of potassic alteration overprinting this sodic alteration. In this case, the nano-scale factor is related to nucleation and growth at the uid-mineral interface, and is controlled not only by external physical parameters such as pressure and temperature, but also by uid composition, in our case the nature of halide (Cl versus F). These dynamic processes remain di cult to assess on a theoretical basis, and experiments remain critical in de ning the mechanisms and kinetics of uid-induced mineral reaction in the crust.
Declarations Figure 1 Fraction of reactant and reaction products as function of time in pure NaCl (A) and NaF solutions (B).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.