Exploration Targeting in Porphyry Cu Systems Using Propylitic Mineral Chemistry: A Case Study of the El Teniente Deposit, Chile

The mineral chemistry of epidote and chlorite from the propylitic halo at El Teniente, in samples collected at distances up to 6.6 km from the deposit center, was determined by microprobe and laser ablation-inductively coupled plasma-mass spectrometry. Results show that both minerals systematically incorporated a range of trace elements that define a much larger footprint to the system than is easily recognized using conventional means such as whole-rock geochemistry. Apart from Fe and Mg in chlorite, there is no significant control of mineral chemistry by bulk-rock composition. For chlorite, geothermometry temperatures and Ti and V concentrations are high proximal, whereas Li, As, Co, Sr, Ca, and Y are low proximal and elevated in distal positions. Ratios of these elements define gradients toward ore varying over three to five orders of magnitude. The proximal-high Ti content is thought to reflect crystallization temperature, whereas proximal-low signatures are believed to characterize elements that are relatively fluid mobile in the inner parts of the propylitic halo in the presence of mildly alkaline to mildly acidic and oxidized fluids so that they are not incorporated into crystallizing chlorite, despite being generally compatible within the mineral structure. These elements begin to substitute into chlorite in the distal parts of the propylitic halo where fluids are largely rock buffered in terms of major element chemistry. In epidote, As defines a broad proximal low and is generally elevated at distances of at least 3 km from the edge of the ore shell. Zinc, La, Yb, Y, and Zr in epidote, among others, appear to define a geochemical shoulder that surrounds the deposit. These patterns are broadly similar to those observed in previous work at Batu Hijau and in the Baguio district, suggesting that these minerals behave consistently in porphyry systems and can therefore provide useful exploration tools within propylitic green rocks. Introduction Porphyry ore deposits represent remarkable accumulations of metals—in particular Cu, Mo, and Au—typically precipitated from hydrothermal fluids in an intrusive host and its surrounding country rocks. Deposits are normally centered within an alteration halo with characteristic mineralogical and chemical zoning patterns (Cooke et al., 2014a). This footprint is a key guide for exploration, providing a larger (several km radius) target area within which mineralization may exist. The most distal alteration facies—the propylitic zone—has been traditionally regarded as a largely isochemical alteration domain (e.g., Sillitoe, 2010) containing little information of use in exploration once beyond the pyrite halo (which may extend into proximal propylitic rocks). Furthermore, the mineral assemblages that characterize propylitic alteration (including actinolite, magnetite, albite, pyrite, epidote, chlorite, sericite, calcite, prehnite, hematite, and zeolites; e.g., Ballantyne, 1981) may be present within barren hydrothermal systems or produced by processes such as regional metamorphism. The lack of research in this environment means that the controls on the formation of these huge alteration zones are incompletely understood. However, because they are so laterally extensive and frequently encountered, an enhanced capability to explore within this domain would be extremely powerful. Indeed, recent work (Cooke et al., 2014b; Wilkinson et al., 2015; Neal et al., 2018; also see Ahmed et al., 2020; Baker et al., 2020; Cooke et al., 2020; Pacey et al., 2020a), part of a series of AMIRA research projects led by the University of Tasmania, has shown that valuable chemical information is locked up in epidote and chlorite that can provide useful exploration insights. Here, we present the results of a study of the propylitic halo of El Teniente, Chile, one of the largest porphyry Cu-Mo systems known, in order to assess whether previously documented patterns are reproduced and to evaluate whether the alteration mineral chemistry footprint extends beyond that defined by conventional whole-rock geochemistry. We consider the differences between the giant El Teniente Cu-Mo system and previous results from the smaller, Cu-Au porphyry systems studied in the Baguio district, Philippines, Batu Hijau, and Indonesia and also from metamorphic background suites. El Teniente Geology and Propylitic Alteration El Teniente is located on the western margin of the Andean Cordillera, approximately 70 km southeast of Santiago, within the confines of the Central Chilean porphyry copper belt. It is the world’s largest underground Cu mine, hosting a premining resource of approximately 95 million tonnes (Mt) of fine Cu (Camus, 2002; Stern et al., 2010), and is the world’s largest known resource of Mo (Sillitoe, 2010) with a current plus mined resource total of approximately 2.5 Mt (Camus, 2002). The deposit is hosted by a late Miocene volcano-plutonic complex, the Teniente Mafic Complex, which occurs within the ©2020 Society of Economic Geologists, Inc. Economic Geology, v. 115, no. 4, pp. 771–791 ISSN 0361-0128; doi:10.5382/econgeo.4738; 21 p. Digital appendices are available in the online Supplements section. 771 †Corresponding author: e-mail, j.wilkinson@nhm.ac.uk © 2020 Gold Open Access: This paper is published under the terms of the CC-BY-NC license. Downloaded from https://pubs.geoscienceworld.org/segweb/economicgeology/article-pdf/115/4/771/5064078/4738_wilkinson_et_al.pdf by guest on 16 June 2020 772 WILKINSON ET AL. Fig. 1. Simplified geologic map of the El Teniente district, showing distribution of major units, prospects, structures, and alteration domains. Area studied for propylitic mineral chemistry shown in the dashed box (see Fig. 2). Modified from Cannell (2004). Abbreviations: act = actinolite, alb = albite, carb = carbonate, chl = chorite, ep = epidote, mag = magnetite, ser = sericite. 36 80 00 36 40 00 36 00 00 6222000


Introduction
Porphyry ore deposits represent remarkable accumulations of metals-in particular Cu, Mo, and Au-typically precipitated from hydrothermal fluids in an intrusive host and its surrounding country rocks. Deposits are normally centered within an alteration halo with characteristic mineralogical and chemical zoning patterns (Cooke et al., 2014a). This footprint is a key guide for exploration, providing a larger (several km radius) target area within which mineralization may exist.
The most distal alteration facies-the propylitic zone-has been traditionally regarded as a largely isochemical alteration domain (e.g., Sillitoe, 2010) containing little information of use in exploration once beyond the pyrite halo (which may extend into proximal propylitic rocks). Furthermore, the mineral assemblages that characterize propylitic alteration (including actinolite, magnetite, albite, pyrite, epidote, chlorite, sericite, calcite, prehnite, hematite, and zeolites; e.g., Ballantyne, 1981) may be present within barren hydrothermal systems or produced by processes such as regional metamorphism. The lack of research in this environment means that the controls on the formation of these huge alteration zones are incompletely understood. However, because they are so laterally extensive and frequently encountered, an enhanced capability to explore within this domain would be extremely powerful. Indeed, recent work (Cooke et al., 2014b;Wilkinson et al., 2015;Neal et al., 2018; also see Ahmed et al., 2020;Baker et al., 2020;Cooke et al., 2020;Pacey et al., 2020a), part of a series of AMIRA research projects led by the University of Tasmania, has shown that valuable chemical information is locked up in epidote and chlorite that can provide useful exploration insights.
Here, we present the results of a study of the propylitic halo of El Teniente, Chile, one of the largest porphyry Cu-Mo systems known, in order to assess whether previously documented patterns are reproduced and to evaluate whether the alteration mineral chemistry footprint extends beyond that defined by conventional whole-rock geochemistry. We consider the differences between the giant El Teniente Cu-Mo system and previous results from the smaller, Cu-Au porphyry systems studied in the Baguio district, Philippines, Batu Hijau, and Indonesia and also from metamorphic background suites.

El Teniente Geology and Propylitic Alteration
El Teniente is located on the western margin of the Andean Cordillera, approximately 70 km southeast of Santiago, within the confines of the Central Chilean porphyry copper belt. It is the world's largest underground Cu mine, hosting a premining resource of approximately 95 million tonnes (Mt) of fine Cu (Camus, 2002;Stern et al., 2010), and is the world's largest known resource of Mo (Sillitoe, 2010) with a current plus mined resource total of approximately 2.5 Mt (Camus, 2002).
The deposit is hosted by a late Miocene volcano-plutonic complex, the Teniente Mafic Complex, which occurs within the Fig. 1. Simplified geologic map of the El Teniente district, showing distribution of major units, prospects, structures, and alteration domains. Area studied for propylitic mineral chemistry shown in the dashed box (see Fig. 2). Modified from Cannell (2004). Abbreviations: act = actinolite, alb = albite, carb = carbonate, chl = chorite, ep = epidote, mag = magnetite, ser = sericite. Agua Amarga Agua Amarga Agua Amarga A g u a A m a r g a F a u l t A g u a A m a r g a F a u l t C o d e g u a F a u l t C o d e g u a F a u l t A g u a A m a r g a f a u l t Moderate propylitic (chl-ep-carb-mag-act) Weak propylitic (chl-carb ± mag-act) Satellite image pale color anomaly Coya-Machali Formation 5 km 5 km 5 km mid-late Miocene Farellones Formation, a >2,500-m-thick sequence of extrusive and intrusive rocks of basaltic to rhyolitic composition ( Fig. 1; Skewes et al., 2002;Cannell et al., 2005).
In the late Miocene-Pliocene (~6.5-4.5 Ma), the Teniente Mafic Complex was intruded by a series of diorite to granodiorite intrusions associated with multiple igneous and hydrothermal breccia complexes. Mineralization developed in concert with three main magmatic episodes (Spencer et al., 2015), each of which was associated with its own pulse of hydrothermal activity (Vry et al., 2010). Thus, it has been concluded that the deposit represents a nested Cu-Mo porphyry system formed by the emplacement and overprinting of a number of porphyry intrusions that evolved separately in space and over time (Astudillo et al., 2010;Vry et al., 2010;Spencer et al., 2015). A number of neighboring hydrothermal centers exist in the district, some of which have distinct surface color anomalies due to argillic alteration at surface (Fig. 1). Olla Blanca, ~6 km to the north-northwest of El Teniente, is the oldest (~9 Ma: Maksaev et al., 2004), and mineralization at La Huifa, ~4 km to the north-northeast, has been dated at ~6.6 to 6.5 Ma (Re-Os molybdenite: Pardo, 2015), slightly predating El Teniente at ~6.3 to 4.6 Ma (Spencer et al., 2015).
Propylitic alteration assemblages are widely developed in the El Teniente district (Fig. 1). North of the Agua Amarga fault, pervasive propylitic assemblages occur in the Coya-Machali Formation and, to a lesser extent, in the Farellones Formation. These assemblages are characterized by pervasive chlorite-epidote-calcite ± hematite, and shallowly dipping carbonate veins (±hematite-epidote halos) are common (Cannell, 2004). The similarity between this alteration and the regional lower greenschist metamorphic assemblage that occurs in Mesozoic and Cenozoic rocks has been noted (Morel, 1984). South of the Agua Amarga and Codegua faults (Fig. 1), the intensity of propylitic alteration of the Farellones Formation decreases significantly. Only weak chlorite-carbonate alteration, largely restricted to amygdules in the volcanic units, has been reported (Cannell, 2004).
Propylitic alteration in the vicinity of El Teniente itself forms a distal halo surrounding a core of potassic alteration that comprises intense biotite-magnetite alteration in the Teniente Mafic Complex or K-feldspar within intermediate-felsic intrusive rocks (Cannell et al., 2005;Vry et al., 2010). The transitional potassic-propylitic zone lies within the 0.5% copper grade contour (Fig. 2) and is characterized by a greenish-brown biotite and an increase of chlorite and paragonitic muscovite in the groundmass. Pyrite ± secondary magnetite abundances are locally elevated, and primary igneous magnetite has been replaced by biotite, chlorite, sericite, anhydrite, and rutile (Cannell, 2004). Sulfides and anhydrite are rimmed by, or intergrown with, chlorite and/or sericite (probably paragonitic muscovite), and plagioclase has been weakly to moderately pervasively altered to muscovite and calcite. Epidote-K-feldspar assemblages have been recorded, particularly in deeper parts of the system near the Sewell diorite (Vry et al., 2010).
Outboard of the 0.5% copper contour, the Teniente Mafic Complex contains between 5 and 20% chlorite, occurring mainly as replacements of hornblende and augite phenocrysts. Pyrite is much more abundant relative to chalcopyrite. Pyrite contains pyrrhotite and chalcopyrite inclusions and is rimmed by chlorite. Green, shreddy biotite constitutes up to 25% of the assemblage and decreases in abundance outward. Accessory magnetite, hematite, ilmenite, rutile, anhydrite, quartz, sporadically developed epidote after plagioclase, and rare phengitic muscovite also occur (Cannell, 2004). Propylitic alteration of intrusive rocks (e.g., Sewell diorite, Teniente dacite porphyry) is typified by chlorite and epidote alteration of the ferromagnesian minerals, plus magnetite, hematite, pyrite, sericite, and calcite (Ossandon, 1974). Veins of quartzepidote (-calcite-chlorite) are variably developed and are cut by quartz-pyrite veins with sericitic alteration halos. Examples of propylitically altered rocks are illustrated in Figure 3.
Previously, this deposit-related alteration was believed to grade out to a district-scale alteration at distances >1 km from the system center (Fig. 1), characterized by replacement of primary mafic minerals and amygdules by chlorite, magnetite, epidote, and hematite, accompanied by weak sericite and albite replacement of plagioclase (Villalobos, 1975;Floody and Huete, 1998).

Methods
One hundred eighty-nine samples of the Farellones Formation (mostly from the Teniente Mafic Complex), comprising largely basaltic to andesitic volcanic rocks and younger intrusions from the El Teniente area, were sampled (Fig. 2 Whole-rock major and trace element data were generated at Acme Analytical Laboratories Ltd. (now Bureau Veritas) in Vancouver, Canada. Samples were jaw crushed to 70% passing 10 mesh (2 mm); a 250-g aliquot was riffle split and pulverized to 95% passing 150 mesh (100 μm) in a mild-steel ring-and-puck mill. A 0.2-g powdered sample was fused in a graphite crucible with 1.5 g of LiBO2/LiB4O7 flux at 980°C for 30 min and then dissolved in 5% HNO3. Major elements were determined using a Jarrel Ash AtomComp Model 975/ Spectro Ciros Vision inductively coupled plasma-emission spectrograph. Trace elements were analyzed using a Perkin-Elmer Elan 6000 or 9000 inductively coupled plasma-mass spectrometer. For major and trace elements, calibration standards, verification standards, and reagent blanks were included in the sample sequence. Reported detection limits are 0.04 wt % for the major elements and 0.5 ppm for the majority of the trace elements, excluding the rare earth elements (REEs; 0.05 ppm). Whole-rock major and trace element data for selected samples are listed in Table 1. Wholerock major and trace element data for all samples are listed in Appendix Table A1.
One hundred thirty-five representative samples containing minerals of interest were prepared as 25-mm polished resin mounts for scanning electron microscopy (SEM) backscattered electron imaging of mineral relationships, electron microprobe-wavelength dispersive spectrometry analysis (EMP-WDS), and laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS).
Major and minor elements in individual chlorite and epidote grains were determined using a Cameca SX100 electron microprobe housed in the Central Science Laboratories at the University of Tasmania. Full results from 755 analyses of chlorite and 316 from epidote are reported in Appendix Tables A2 and A3, respectively. Major, minor, and trace elements in the same grains were measured using a New Wave 193nm solid-state laser coupled to an Agilent 7500cs quadrupole mass spectrometer, located in the School of Physical Sciences, Discipline of Earth Sciences, University of Tasmania. Typically, five to 10 spot analyses were acquired from each sample, from within three to five separate mineral grains. For each analysis, 30 s of background signal was acquired prior to 60 s of mineral ablation. Ablation was performed in an atmosphere of pure He (~0.7 L/min) using spots ranging in diameter from 30 to 55 μm with a laser repetition rate of 10 Hz. Laser beam fluence at the sample was ~3 J/cm 2 .
For epidote, a total of 37 elements were determined on each spot by analysis of the following isotopes: 23 Na, 24 209 Bi, and 238 U. Acquisition time for each mass was set to 0.02 s, with the exceptions of Na, Mg, and Si (0.005), Al, K, Ca, Mn, Fe, and U (0.01), and Co, Cu, As, Ag, Sn, Sb, and Bi (0.03).
No corrections for interfering molecular or isobaric species were required due to the selection of appropriate analytical isotopes and the minimization of molecular species formation. Signal processing was done using an in-house Excel macro and the ExLAM Excel VBA application (Zacharias and Wilkinson, 2007). In total, 641 LA-ICP-MS spot analyses of chlorite (from 115 samples) and 592 of epidote (from 103 samples) meeting quality control criteria (i.e., avoiding contamination from mineral inclusions) were obtained.
Aluminum and calcium concentrations determined by microprobe were used as the internal standards for reduction of chlorite and epidote LA-ICP-MS data, respectively. NIST612 standard reference material was used for external calibration of the LA-ICP-MS results according to standard practice. Full LA-ICP-MS results are reported in Appendix Table A4.

Whole-rock geochemistry
Key pathfinder elements were grouped into eight percentile bins (1, 5, 10, 25, 75, 90, 95, and 99%) and plotted in plan view to define the conventional geochemical footprint of the deposit. Anomalous Cu above a background population (>230 ppm), defined by a natural break in the cumulative probability curve, is observed in a fairly coherent anomaly surrounding the deposit, extending up to 1 km from the edge of the 0.5 wt % Cu contour. Some additional anomalous samples are observed between El Teniente and La Huifa but none in the vicinity of Olla Blanca (Fig. 4A).
Anomalous Mo (>2.1 ppm) is somewhat erratically distributed across the district. Consistently elevated values are observed close to El Teniente but are very proximal, occurring within the 0.5 wt % Cu contour. As with Cu, there are elevated values in the vicinity of La Huifa, but for Mo there is a marked gap between the two systems, with background values in between (Fig. 4B). Olla Blanca samples are not anomalous in Mo.
Anomalous Mn (>1,280 ppm) is more broadly distributed than the principal ore metals and appears to form a crude halo around the 0.5 wt % Cu ore shell, with maximum values between 0.5 and 1 km from its edge and lower concentrations within it (Fig. 4C). Neither La Huifa nor Olla Blanca show Mn anomalism.
The Zn anomaly around El Teniente (>170 ppm) is slightly better developed than Mn, but they are comparable in form (Fig. 4D). Their covariance is confirmed by a robust linear regression that yields an R 2 value of 0.74. One point to note for both Mn and Zn is that the results for neighboring samples, or for those from the same drill hole, can be very variable, indicating a nugget effect on the hand-sample scale. As with Mn, neither La Huifa nor Olla Blanca show enrichment in Zn.
Lead shows a similar pattern to Cu with a slight asymmetry to the north and west but has a greater extent of anomalism (above 12.9 ppm) up to 2.5 km from the edge of the 0.5 wt % Cu contour (Fig. 4G). There is arguably a small separation between El Teniente and La Huifa, with the latter system showing some apparently associated anomalous samples to the south and east, very much like Cu and Mo. Olla Blanca is not anomalous in whole-rock Pb (Fig. 4G).
It is difficult to define the background level for As in the district, because ~90% of the data fall above the upper crust-   (Rudnick and Gao, 2003) suggesting a very broad enrichment that may be partly premineralization in origin based on the elevated values in ostensibly unaltered samples (Table 1). Using the same percentile bins as above, a relatively weak footprint is defined (Fig. 4E) with elevated values (>31 ppm) occurring in a halo around El Teniente at a distance of 0.5 to 1 km from the edge of the ore shell. La Huifa shows no As anomalism; conversely, Olla Blanca has markedly elevated As concentrations in a number of samples. Alternatively, if a complex set of natural breaks in the data are considered, values above a putative background of 15 ppm define a very broad, asymmetric footprint that extends up to 7.5 km from the edge of the 0.5 wt % Cu contour in the northwest quadrant and to the limit of sampling in most areas (Fig. 4F). The most clearly defined limit is at around the 6225000 northing, to the south of which only low As concentrations are observed. Several elements show proximal low patterns that can be discerned in whole-rock data, most notably Na and Sr. For Na, based on natural breaks in the data, anomalously low concentrations below 2.45 wt % fairly effectively define a proximal footprint broadly centered on the orebody but extending up to 1.5 km from the 0.5 wt % Cu contour to the north and west (Fig. 4H).
In all cases, the whole-rock anomalies defined are imperfect in the sense that the samples with elevated values in the inferred footprints are frequently accompanied by neighboring samples with background-level concentrations. This implies heterogeneity in pathfinder metal distribution that may reflect control by somewhat erratically distributed, perhaps structurally controlled, sulfides.

Chlorite major element chemistry and thermometry
Major element compositions of chlorite correspond to the clinochlore-chamosite solid solution with a wide range in Fe (8.4-26.6 wt %) and Mg (5.1-15.2 wt %) content determined by microprobe, excluding one very Fe rich sample (Fig. 5) The limit of detection where concentrations could not be determined is denoted by "<xx"; all major oxides, total C, S, and loss on ignition (LOI) in wt %; trace elements in ppm apart from Au, Pd, Pt, and Re, which are in ppb; N = number of analyses included in average; n.d. = not determined; representative samples of different rock types were used, screened for alteration effects, such as elevated LOI (>2.5 wt %) or ore metals, and excluding outlying compositions for a given rock type; full data set reported in Appendix Table A1 10.2 and 13.2 wt %, respectively. Manganese contents determined by microprobe ranged up to 6.1 wt %, but the vast majority were <2 wt % Mn. Chlorite crystallization temperatures were calculated from microprobe data using the thermodynamic model of Walshe (1986). Sample average temperatures derived from this approach are in the range 113° to 356°C with an overall mean of 241°C (App . Table A2). Gridding the data shows that the temperatures define a clear thermal anomaly associated with the ore deposit and with a subsidiary anomaly in the vicinity of La Huifa (Fig. 6).
Although chlorite thermometry is complex, with the temperature-related Al IV Tschermak substitution (Si 4+ ) IV (Mg 2+ ) VI = (Al 3+ ) IV (Al 3+ ) VI also related to aH 4 SiO 4 , aAl3 + /(aH + ) 3 , aFe 2+ / (aH + ) 2 , aMg 2+ /(aH + ) 2 , and aH 2 O (de Caritat et al., 1993), we can argue that, at least in a semiquantitative sense, the thermometry results are robust. This is because silica activity is likely to be buffered at quartz saturation in propylitic fluids, and pH and Al activity will be buffered by the aluminosilicate assemblage in the host rocks at the relatively low water/ rock ratios expected in the propylitic domain. We also know from fluid inclusion data that propylitic fluids at El Teniente are water rich (Wilkinson et al., 2019) and that aCO 2 must be low for epidote to be stable (Bird and Spieler, 2004) so that we can assume aH 2 O ~1.0. The principal remaining variables that may have some effect are the activities of Mg 2+ and Fe 2+ in the fluid. These may be controlled, at least in part, by the composition of the host rock, with more mafic compositions yielding higher Fe chlorites (see below). However, the relatively uniform bulk composition of most of the host rocks (104 out of 189 are basaltic andesite-andesite based on silica content; App. Table A1) means that any effect will be limited. Finally, we see an excellent negative correlation between chlorite Si IV and calculated temperature (robust regression R 2 = 0.77), consistent with the Tschermak exchange and as observed in geothermal data sets (e.g., Cathelineau, 1988). Thus, although the absolute temperatures reported (App.

Chlorite trace element chemistry
Manganese, Ca, Zn, V, K, Co, Ni, Ti, Li, Na, and Ga were the most common trace elements (generally 10s to 100s of ppm) determined in chlorite in decreasing order of abundance (Fig.  7). The presence of Mn, Zn, Ni, and Li is not unexpected, as they represent partial solid solution toward pennantite, baileychlore, nimite, and cookeite end members, respectively, of the chlorite series. Vanadium 3+ may substitute for Al 3+ in octahedral coordination, analogous to Cr 3+ in kämmererite, and Co 2+ has similar ionic radius to Fe 2+ and Ni 2+ , so it is likely to occur in similar octahedral coordination. Titanium 4+ is also known to substitute into octahedral sites.
The trace components Cu, Sr, As, Ba, Sn, Pb, Y, Zr, Ce, and La, mostly present at concentrations of tens of parts per million or less, were also detected in more than two-thirds of the analyses (Fig. 7) and are assumed to be substituted into chlorite structural sites on the basis of smooth signals in LA-ICP-MS analyses. Copper shows a particularly wide range and a positive skew, which may reflect the unrecognized incorporation of sulfide nanoinclusions in some analyses; however, many high signals are extremely homogeneous and consistent from grain to grain in a given sample so that, for the most part, high-Cu chlorites probably do reflect structural substitution. Close correspondence between the ionic radius of Cu + and Li + supports their potential substitution in the same sites. Less consistently detected were Cr, B, Mo, Sb, Ag, Bi, and U, which may reflect their presence in mineral nanoinclusions. The lowest limits of detection for some of the heavier elements were ~4 ppb.

Protolith controls of chlorite chemistry
An important factor to consider when interpreting mineral chemistry data is the possible control by the protolith (bulkrock chemistry) on major and trace element compositions.

Mn Ca Zn V K Co Ni Ti Li Na Ga Cr Cu B Sr As Ba Sn Pb Mo Sb Y Zr Ce La Ag Bi U
However, it is often difficult to separate protolith from spatial controls, particularly in systems where lithologies are heterogeneously distributed. This is the case at El Teniente, where more felsic intrusive rocks predominate in the central parts of the system (Fig. 1). In order to evaluate protolith effects, the data reported here were divided into groups based on host rock lithology (see App. Table A1). This approach shows that there is a likely influence on Fe and Mg incorporation in chlorite (App. Fig. A1A), with more mafic rocks hosting more Fe-rich chlorite. Because of the strong coupled substitution with Mg, this also leads to lower Mg in chlorite in these samples, even though these rocks are still relatively Mg rich. Silicon and Al show no resolvable protolith control. Most trace elements in chlorite show no correlation with lithology, with the exception of some of those shown in Appendix Figure A1B. Arsenic and Sb tend toward higher concentrations in the more mafic rocks, but this is probably related to their tendency to be located in more distal positions. This behavior is more clearly illustrated by K and Zr, which are rich in the more felsic rocks but are generally elevated in chlorite from the more mafic lithologies, probably for the same reason (enriched in more distal propylitic samples). Nickel and Cr can display protolith effects, but there is no evidence of this in the El Teniente data.

Chlorite spatial variations
The major elements do not show very systematic spatial patterns. Iron and Mg are influenced by protolith as noted above, which obscures spatial patterns, but there is possibly a high shoulder in the Fe data located outside the ore shell (see App. Fig. A2), similar to what was observed at Batu Hijau . Aluminum perhaps displays a tendency toward higher concentrations in proximal positions (indirectly reflecting a higher formation temperature) and Si tends to be low in proximal samples (see App. Fig. A2).
Clear spatial zonation patterns are observed for a number of trace elements in chlorite, including a proximal high for Ti, with additional small anomalies spatially associated with La Huifa and Olla Blanca (Fig. 8). Vanadium shows a proximal shoulder, with lows in the core of the deposit and a maximum around the margin of the copper ore shell (Fig.  8). There are proximal lows for Li, As, Co, Sr, Ca, La, and Y, and these elements increase distally (Fig. 9). Similar patterns of trace element variation in chlorite have been noted at Batu Hijau and Resolution Cooke et al., 2020).

Epidote major element chemistry
Major element compositions of the analyzed epidote correspond to true epidote, with a systematic inverse correlation between Fe and Al representing compositions along the epidote-clinozoisite solid solution (Fig. 10). Iron concentrations were mostly in the range 6.5 to 13.1 wt % (median 10.0 wt %) and Al in the range 10 to 15 wt % (median 13.3 wt %). The median Ca and Si contents were 16.5 and 17.3 wt %, respectively. Manganese contents determined by microprobe ranged up to 2.4 wt %, but the vast majority were <1 wt %.

Epidote trace element chemistry
Manganese, Sr, Mg, Ti, V, and As were the most common trace components in epidote (in decreasing order of abundance), with concentrations mostly exceeding 100 ppm (Fig.  11). Substitution of Mn, Sr, and V represent the solid-solution series toward piemontite, epidote-Sr, and mukhinite, respectively. Titanium and Mg were also frequently detected by microprobe (also see Cooke et al., 2014b;Baker et al., 2020), which, with its much better spatial resolution (~5 vs. 40 µm), is more likely to avoid unrecognized mineral contaminants so that these elements are likely to be structurally bound. Zonation in LA-ICP-MS maps (Cooke et al., 2014b shows that As (and Sb) is also incorporated into the epidote structure, probably as As 3+ in the M site.
Trace components Ga, Pb, Na, Sb, Zn, Ce, La, Y, Zr, Sn, Ba, Eu, U, Yb, Th, and Lu were determined in the 0.01to 200-ppm range in more than two-thirds of the analyses (Fig. 11) and are also believed to be substituted into the Hf, Au, Tl, and Ta were also detected but less consistently and may be present in mineral nanoinclusions that were too small to be discriminated as separate phases during laser signal processing. For K (Na), Cu, and Bi, this interpretation is supported by the marked positive skew observed in the results, with markedly higher mean than median concentrations (Fig. 11). The particularly wide ranges observed for Sb, the REE-Y, Th, and U is due to compositional zoning that is frequently observed in SEM-backscattered imaging and in LA-ICP-MS maps (Cooke et al. 2014b.

Protolith controls of epidote chemistry
For epidote, there are no discernible effects of protolith on epidote major element chemistry (App. Fig. A3A). For the trace elements, the vast majority show no discernible relationship to lithology, perhaps with the exception of Ba, Pb, Sr, and Zr (App. Fig. A3B). Barium appears to be slightly richer in basalts relative to andesites and monzonites; however, the granodiorite-hosted samples tend to be slightly enriched.  11. Summary of compositional ranges for minor and trace elements in epidote from El Teniente. Dot represents mean value; horizontal line represents median; box represents 25 th to 75 th percentiles, whiskers represent maximum and minimum excluding outliers; circles represent ouliers (>1.5× interquartile range from the box); triangles represent far outliers (>3× interquartile range from the box).

Mn Sr Mg Ti V As Ga Pb Na Sb K Zn Ce La Y Zr Sn Ba Eu Cu U Yb Co Mo Bi Hf Th Lu Au Tl Ta
Lead is variable and probably responds more strongly to spatial position, but epidote is possibly slightly enriched in lead in more Pb-rich granodiorite samples. Epidote in the more felsic samples is generally enriched in strontium but, because of the spatial distribution of lithologies, it is difficult to separate this from any spatial effects. This is illustrated by Zr, which is relatively low in epidote in the more felsic (Zr-rich) host rocks and therefore may reflect a predominant spatial control with proximal epidotes showing depletion in Zr (Fig. 12), similar to its behavior in the Collahuasi district .

Epidote spatial variations
For the major elements, only Fe shows any systematic spatial pattern, forming a proximal shoulder with a maximum around the edge of the ore shell, somewhat similar to V in chlorite (App. Fig. A4). Systematic patterns are observed for a number of trace elements in epidote, including Cu, As, La, Yb, Zn, Y, and Zr (Figs. 12, 13). A prominent, proximal, positive Cu anomaly overlaps with the ore shell, comparable to the pattern around the Rosario porphyry ), but appears to be offset to the southeast. A second Cu anomaly is located to the west-southwest, related to localized mineralization intercepted in one drill core in the Agua Amarga area that also had elevated trace elements in whole rock (e.g., Fig. 4B-E). Arsenic displays the most coherent halo (also observed at Ujina; see Baker et al., 2020), with a very broad, butterflyshaped, proximal low that is offset slightly to the north and west of the orebody and extends for >3 km from the edge of the ore shell. At the limits of sampling there is a suggestion that a distal high is developed. Halo-type anomalies appear to be present for the other elements shown, with peak concentrations reached about 1 to 2 km from the edge of the orebody and with the addition, in the case of Zn and perhaps the REEs, of a small proximal anomaly.

Discussion
A number of mineral chemistry features identified in this study display gradients away from the core of the El Teniente hydrothermal system, as defined by the position of the ore shell. When considering such relationships, it is important to bear in mind the third dimension; on average, the surface samples collected are ~800 m above the level of the center of the deposit as currently mined (Teniente-8, the deepest level is at 1,980 m above sea level) and some are nearly 1,600 m higher, so there is a significant vertical component. Furthermore, the deposit has been tilted, postmineralization, by ~10° to the west (Vry et al., 2010), so that a slight westward offset of the overlying propylitic anomaly might be anticipated. This could be further exaggerated by the topography. Thus, on the west side, the proximal propylitic zone might extend farther, and propylitic grade might be lower on the west side for a given elevation (Fig. 14).

Chlorite vectoring and fertility indicators
Results from chlorite geothermometry confirm that El Teniente sits within a strong thermal anomaly that can be resolved for approximately 2 to 2.5 km beyond the edge of the 0.5 wt % Cu grade contour. This is comparable to that previously observed for Batu Hijau . As with Batu Hijau, it is noteworthy that the thermal anomaly closely maps the form of the ore deposit itself and is asymmetric in the surrounding propylitic rocks. In the case of El Teniente, there is a steeper gradient to the south. We suggest that this southern boundary is modified by the presence of the Teniente fault zone and the southern boundary of the Sewell diorite, both of which are parallel to the Agua Amarga fault (Fig. 1). Thus, this may reflect a postmineralization offset that has down-dropped the southern block, thereby concealing more proximal propylitic signatures at depth. The data also resolve a second thermal anomaly to the northeast that corresponds to the position of the La Huifa deposit, with lower thermal gradients developed in the area between La Huifa and El Teniente (Fig. 6).
Chlorite is also enriched in Ti in a bullseye-type anomaly that is coincident with the thermal anomaly (Figs. 6,8), consistent with the conclusion from Batu Hijau that Ti substitution into chlorite is thermally controlled . The Ti content of chlorite is a more sensitive vector toward the porphyry center than temperature, because it varies over two orders of magnitude (6-650 ppm). It also defines a gradient that seems to extend out farther laterally, between 1.5 to 2.5 km beyond the upward vertical projection of the ore shell, depending on direction (Fig. 8). There is a distinct westward expansion of the Ti anomaly that could reflect the geometric and geologic factors noted above. Comparable, proximal enrichment of Ti in chlorite is also noted at the Resolution porphyry Cu-Mo deposit .
Vanadium in chlorite also shows a proximal anomaly but forms more of a geochemical shoulder that sits on the fringe of the ore shell (Fig. 8)-behavior that is also very similar to Batu Hijau. Outside this shoulder, V gradients extend at least 2.5 km beyond the upward vertical projection of the edge of the ore shell, with the greatest extension again to the west. Although care must be taken not to overinterpret the gridded data, the V map suggests the presence of a west-northwest structural corridor, marked by a pronounced concentration low, be-  1) and is thought to have played an important role in controlling the emplacement of magmas and hydrothermal activity in the deposit, as well as being a long-lived basin-bounding structure that is still active (e.g., O. Rivera and M.F. Falcón, unpub. report, 1998). The Codegua fault is one of many such arc-transverse structures developed along the Andean margin that are thought to reflect reactivation of a pre-Andean basement fabric (O. Rivera and M.F. Falcón, unpub. report, 1998).
A number of elements show proximal lows, with Li being a particularly good example (Fig. 9). This is attributed to mobility of Li (together with Co, Sr, Ca, and REE-Y) in high-temperature potassic and mildly acidic proximal propylitic fluids, as has been constrained by chemical mass transfer analysis in the Northparkes Cu-Au porphyry cluster, New South Wales, Australia (Pacey, 2016). Very similar patterns have also been observed at Batu Hijau  and at Resolution .
As vectoring tools, element ratios in chlorite (proximitor ratios; Wilkinson et al., 2015) are particularly effective because they can enhance significantly the signal gradient around a porphyry center and provide a better discrimination from background. Useful ratios identified at Batu Hijau are shown in Figure 15 for El Teniente, with the addition of V/Li. These ratios vary over three to five orders of magnitude and clearly define the mineralized center of the system. The Ti/Li anomaly is rather remarkable in that it precisely maps the ore shell, including its triangular form, and extends at least 3 km from the edge of the ore shell. This ratio was also very effective in defining the center of the Resolution porphyry system . All of the Ti-based ratios, with the exception of Ti/K, clearly coincide with the system center, although Ti/Ni shows a proximal shoulder with the maxima aligning with the 0.5 wt % Cu grade contour. Similar behavior is observed for V/Li. The V/Ni ratio displays an intermediate halo with maxima developed just outside the ore shell to the east but shifted 1 to 2 km away from it in the west and northwest. The distal expression of this pattern arguably extends 5 km from the ore shell in this direction; the same ratio was considered to define one of the broadest halos at Batu Hijau, extending ~5 km from the center of the deposit .
Significantly, all the Ti-based ratios also show positive anomalies associated with the neighboring alteration-mineralization systems of La Huifa (to the north-northeast) and Olla Blanca (to the north-northwest), with the anomalies appearing to define a connected, lower-intensity corridor. This is similar to the trend noted at Batu Hijau between the deposit and the Sekongkang prospect, interpreted to reflect the geometry of the underlying parental batholith . For El Teniente, although the propylitic alteration system appears connected between the prospects and deposit, the pattern must be composite, built from overlapping events, given the difference in age of the neighboring systems (Olla Blanca, ~9 Ma; La Huifa, 6.6-6.5 Ma; El Teniente, ~6.3-4.6 Ma).
Plotting the El Teniente data as a function of radial distance from the deposit center shows the same general trends of decreasing proximitor ratios with distance that were observed at Batu Hijau, as exemplified by the Ti/Sr ratio (Fig. 16). A robust regression of the data defines a proximitor equation that predicts distance from the center of El Teniente (X) to an accuracy of ±640 m (1σ): This equation has a slight negative bias, particularly at greater distances, where it tends to slightly underestimate distance to center. Data derived from samples proximal to the La Huifa and Olla Blanca centers plot off this trend, as would be expected if these samples reflect the local anomalies associated with these centers.
The specific characteristics of the chlorite anomalies associated with the two prospects differ in a number of ways from those developed around El Teniente itself (e.g., La Huifa has a proximal V/Ni anomaly, perhaps due to its small dimensions, and both of the prospects are subdued with respect to the Li-based ratios), likely due to different exhumation levels and geologic environments. Olla Blanca is interpreted as an eroded caldera with extensive silification and hot spring activity surrounded by propylitic alteration but with porphyritic intrusions and strong propylitic alteration at depth (Camus, 1977). La Huifa is a  mineralized tourmaline breccia, the Extravío breccia, potentially linked to a porphyry system at depth. From an exploration standpoint, it is significant that the El Teniente anomaly differs significantly from the subeconomic prospects. The Mg-bearing ratios proposed for Batu Hijau (Mg/Ca and Mg/Sr) do not clearly define the system at El Teniente, probably because of the influence of protolith composition, which can influence the Mg and Fe content of chlorite: very proximal samples at El Teniente are dominated by intermediate to felsic intrusive lithologies as opposed to the basaltic to andesitic composition of the majority of the remainder (App .  Table A1).
In terms of fertility assessment, the Mn and Zn contents of chlorite appear to provide a useful guide, being generally rich in porphyry systems. Comparison of the El Teniente data with results from the Baguio district and several background data sets, including some regional metamorphic terranes (Fig. 17), shows that although there is a lot of overlap at low concentration ranges (which tend to be chlorite data from very proximal or very distal locations relative to Fig. 16. Plot of the natural log of the Ti/Sr ratio in chlorite as a function of radial distance from the center of El Teniente compared with results from Batu Hijau . As expected, data from samples proximal to La Huifa and Olla Blanca (shown in yellow and blue) do not follow the trend. Robust linear regression of the El Teniente data (red) produces a reasonable regression fit (R 2 = 0.69) that defines the relationship between Ti/Sr ratio and distance. A new, robust regression of the Batu Hijau data ; data in the range 764-2,300 m, excluding data from the west traverse) gives a good fit (R 2 = 0.84) but with a much steeper slope.

Baguio district
Metamorphic the porphyry centers), it is almost exclusively the porphyry systems that have elevated values, above ~4,000 ppm for Mn and 900 ppm for Zn (also see data from Resolution; Cooke et al., 2020).

Epidote vectoring and fertility indicators
Epidote trace element patterns generally define a broader but less well-defined footprint than chlorite. Arsenic is the best vector element, displaying a broad low in proximity to ore, with a distal enrichment to at least 5 km from the system center, especially to the north and west (Fig. 12). Other elements appear to define a geochemical shoulder, or halo, with maxima at varying distances from the center, but typically at 1 to 2 km from the edge of the ore shell. These patterns are broadly similar to those identified in the Baguio (Cooke et al., 2014b) and Collahusai  districts. It is noteworthy that the features identified in chlorite that appear to correspond to the La Huifa and Olla Blanca systems are not so evident in the gridded epidote data (Figs. 12, 13). The most diagnostic features of El Teniente epidote chemistry in relationship to background metamorphic rocks and smaller deposits, such as those of the Baguio district in the Philippines, are the significantly elevated As and Sb concentrations, as well as a general depletion in heavy REEs (Fig. 18). It is noteworthy that the maximum As and Sb concentrations reached in the El Teniente district are broadly similar to, but extend to higher values than, those reported from the Collahuasi district . Such signatures may therefore be an indicator of super-giant systems, although the possibility of anomalous As and Sb addition in continental arc porphyry systems as opposed to oceanic arcs cannot be excluded, so this relationship may not extend to Cu-Au porphyries developed in island arcs. Nonetheless, it is clear that the significant fluxing of such metals in porphyry-related magmatic-hydrothermal El Teniente Baguio district Georgetown inlier Harts Range West Scotland Fig. 18. Plot of Sb and Yb versus As concentrations in epidote from the El Teniente area in comparison to data from the Baguio porphyry district, Philippines, and several regional metamorphic terranes. Georgetown inlier and Harts Range Proterozoic terrane data from Baker et al. (2017), Baguio data from Cooke et al. (2014b), and West Scotland Paleozoic Moine and Dalradian data from Wilkinson et al. (2017). Note that a large proportion of metamorphic analyses (not shown) were below the limit of detection for As and Sb (typically ~1 ppm for As and 0.1 ppm for Sb). systems makes the discrimination from regional metamorphic epidote fairly straightforward.

Conclusions
Chlorite and epidote mineral chemistry data from El Teniente support the contention that systematic patterns in propylitic mineral chemistry are developed around porphyry systems that extend up to 5 km beyond the orebody. A range of elements show low concentrations in proximal samples (Li, As, Co, Sr, Ca, and REE-Y in chlorite; As, La, Yb, Zn, Y, and Zr in epidote) that increase toward the fringes of the zone of hydrothermal influence or increase to an intermediate position to form a halo and then decrease to the edge of the porphyryrelated propylitic zone. The former are generally effective for discriminating propylitic alteration from regional background signatures, because they are significantly elevated above typical background at the limits of propylitic alteration halos. A number of other elements are high proximal and decrease outward (Ti and V in chlorite; Fe and Cu in epidote). Ratioing proximal high elements to proximal low elements yields values that vary over several orders of magnitude and these define steep gradients toward ore. Not only do mineral chemistry indicators display more robust footprints than conventional whole-rock data (less prone to nugget effects), they also show much better defined and laterally extensive gradients toward ore. These patterns arguably demonstrate a much wider influence of magmatic-hydrothermal fluids in porphyry-related propylitic halos than previously considered, challenging the paradigm that these alteration domains are formed from convecting groundwater. This conclusion is further supported by chemical mass transfer and isotopic data (see Pacey, 2016;Pacey et al., 2020b).