Abstract
We present 40Ar/39Ar ages of dacite domes and rare volcanic sanidine megacrysts from Taápaca volcano (northern Chile) that record 1.3 m.y. of activity. Our focus is on 20 megacrysts from a single 32.9 ka eruption. We interpret that their surprisingly correlated Ba-rich and Ba-poor growth zones separated by resorption surfaces reflect frequent heat pulses with a uniform thermal history over >300 k.y. of growth. We infer extended storage in small (<400 m), shallow “hot” domains within a larger magma system. Our findings bear on the origin of K-feldspar megacrysts in plutonic rocks, thus linking volcanic and plutonic processes in shallow silicic magma systems, and support protracted residence of hot magma in small batches at upper-crustal levels to produce megacryst-bearing granitoid intrusive complexes.
INTRODUCTION
Emplacement, storage, and mobilization of silicic magmas control their eruption (Lipman and Bachmann, 2015; Cashman et al., 2017) and formation of silicic intrusive bodies (Keller et al., 2015). However, in parallel with the contrasting models for silicic intrusions (e.g., Johnson and Glazner, 2010; Gajos et al., 2016; Müntener et al., 2018), the depths at which silicic magmas originate are heavily debated. Although silicic eruptions are generally inferred to be derived from shallow upper-crustal hot reservoirs (Deering and Bachmann, 2010; Gelman et al., 2014; Lee and Morton, 2015; Andersen et al., 2019), Glazner et al. (2004), Annen (2009), and Glazner (2021) have instead argued for a lower-crustal origin of silicic magmas with only transient existence of melt-rich bodies in the upper crust. Therefore, constraints on the nature and size of the magma bodies as well as the frequency of recharge events are essential to understand the incubation and mobilization of crustal reservoirs and their eruptions.
We present diffusion and thermal modeling on 20 sanidine megacrysts from a single dacite dome eruption of Taápaca volcanic complex (TVC; 5824 m above sea level, 18°S, 69°W) in the Central Andes. Our results place tight constraint on the nature and frequency “ramp-up” of recharge events into the crystals’ parent reservoir, support the model of shallow above-solidus “hot” storage, and address questions surrounding the link between volcanic and plutonic records of igneous processes (e.g., Bachmann et al., 2007; Gelman et al., 2014; Glazner et al., 2015; Schaen et al., 2021).
TVC evolved in four stages (Clavero et al., 2004) for >1 m.y. and produced uniform dacites with mafic enclaves and 20–30 vol% crystals, including unusually abundant (1–7 vol%) and large sanidine megacrysts (1–12 cm in size; Fig. 1; for geological background and sample descriptions, see the Supplemental Material1). Rout et al. (2021) studied Ba-zonation patterns in such megacrysts from stage II to IV, indicating that they initially formed in rhyodacite magmas and were later erupted in hybrid dacites after mafic recharge.
Here, we analyzed 20 megacrysts (up to 8 cm; Or65–70Ab30–35An<1, where Or is orthoclase, Ab is albite, and An is anorthite) from a dacite dome TAP-97-28 (18°07′S, 69°31′W) from stage IV and constrain the nature and pre-eruptive history of the reservoir. Methods and details of electron probe microanalysis and 40Ar/39Ar dating are provided in Supplemental Text and Datasets S1–S3.
RESULTS
Sanidine crystals from two rhyolite dome lavas (TAP-97-40 and TAP-97-41) 4.5 km to the NE of the TVC summit give 40Ar/39Ar ages of 2.18 ± 0.06 and 3.76 ± 0.09 Ma (Table 1). However, the temporal gap between these and the oldest TVC dacite eruptions (1.27 ± 0.04 Ma; Wörner et al., 2000) and the distinct composition of the dacites indicate that the TVC sanidine crystals originate from an emergent magmatic system unrelated to earlier reservoirs. The 40Ar/39Ar ages (Table 1) of sanidine from 10 TVC lavas document the long eruptive history of TVC between 979 ± 280 ka and 32.1 ± 5.0 ka (±2σ analytical uncertainty). Holocene ages for dome-collapse breccias (Clavero et al., 2004) are in accord with block-and-ash flows deposited during the last glacial stage. The total duration of dacitic eruptions at TVC thus covers the past 1.3 m.y., with recent eruptions implying an active magmatic system prone to future eruptions.
Our focus was on TAP-97-28, one of the youngest dated domes (32.9 ± 3.6 ka). Sanidine megacrysts (Fig. 2; Supplemental Dataset S2) in this dome show correlated BaO-zonation patterns: Cores are BaO poor (0.2–1 wt%), followed by a resorption boundary and a BaO-rich (1.5–2.4 wt%) inner mantle. This mantle (mantle 1) shows frequent resorption interfaces and (re)growth zones with a sharp increase in BaO by up to 1.5 wt%. BaO then smoothly decreases to <1 wt%. The BaO jump at the resorption interfaces roughly correlates to their penetration depth (Text S1). Outward from the inner mantle, we observe low BaO (<0.6 wt%) oscillatory growth followed by an outer mantle (mantle 2) with ~2.2 wt% BaO. Most crystals have a thin (1–2 mm) outermost BaO-rich (2–2.4 wt%) rim with high-frequency BaO oscillations.
Except for Ba, sanidine compositions are largely uniform from core to rim (Fig. 2) without evidence of chemical changes in the parent rhyodacite magma. Using amphibole inclusions in the megacrysts (Fig. 2), we estimated temperatures of 720–827 °C and shallow pressures of 0.1–0.2 GPa (after Ridolfi, 2021; Dataset S4). Rhyolite-MELTS (v.1.2.0; Gualda et al. 2012) modeling of the parent rhyodacite composition (Blum-Oeste and Wörner, 2016) suggests sanidine growth at 720–770 °C and >40% crystallinity (Dataset S4), consistent with amphibole crystallization temperatures of 720–827 °C. Dissolution of Ba-rich sanidine at high temperature (>770 °C) led to a Ba-rich melt boundary layer (Ba being highly compatible in sanidine; GERM KD Database, https://earthref.org/KDD/). Regrowth from this boundary layer following each dissolution event started with a sharp Ba increase (Fig. 3E). Further regrowth gradually depleted the boundary layer and resulted in decreasing Ba. This type of “sawtooth” Ba zonation thus directly reflects repeated heat pulses and cooling in a shallow magma reservoir. The lack of chemical changes (except Ba) in the parent melt suggests that recharge events causing the heat pulses did not mix or hybridize the resident rhyodacite magma, which was “locked up” due to its high crystal content. Although mixing with a similar rhyodacite recharge is also an option, a much hotter (i.e., “superheated”) crystal-poor rhyodacite with the exact same composition as a >40% crystallized rhyodacite would be required, which is unlikely. Only a final, critical mafic recharge of sufficient intensity or proximity to the rhyodacite resulted in mixing and hybridization, as is evident from the sharp compositional changes toward the crystal rims (Fig. 2) and frequent mafic enclaves in the (hybrid) dacite.
We apply a non-isothermal diffusion model (Petrone et al., 2016; Rout et al., 2020) to the Ba profiles across the resorption boundaries. We estimate the best-fitting “Dt value” (D = diffusion coefficient, t = time), i.e., square of diffusion length √Dt, reflecting the extent of diffusion without considering temperature and its uncertainties. In all megacrysts, Dt values fall between 1 and 60 µm2 (Dataset S5) and decrease from core to rim. Comparing Dt at equivalent zonal positions in different crystals, we observe values of 20–35, 4–7, and 0.4–2 µm2 for core–inner mantle, oscillatory zone–outer mantle, and outer mantle–rim boundaries, respectively (Fig. 3A).
To constrain the timing of the heat pulses, we calculate diffusion times for each zone boundary (Fig. 3; Dataset S5) using zone-specific amphibole temperatures. We calculate partial diffusion times at each boundary, i.e., from the onset of its diffusion until the next younger boundary formed, including the resorption time in between (Supplemental Text S1). The partial diffusion times become shorter from crystal cores (90 ± 29 to 170 ± 70 k.y.) to rims (0.3 ± 0.1 to 2.5 ± 0.6 k.y.) with accumulated diffusion (and residence) times up to 375 ± 120 k.y. However, we note that the actual residence times could be longer, as the earliest modeled diffusion starts after the first zone boundary formed and not at the onset of crystallization. The 40Ar/39Ar dating of four slices taken from the core, mantle, and rim of megacryst TAP-97-28-1 (Table 1) yield ages of 33.4 ± 3.6 to 38.6 ± 5.2 ka.
DISCUSSION
The intracrystal 40Ar/39Ar ages of TAP-97-28-1 are identical within uncertainty of the 32.9 ± 3.6 ka eruption age of its dacite matrix and indicate efficient diffusive loss of radiogenic Ar from the megacryst during the long (375 ± 120 k.y.) storage. The short pre-eruptive diffusion times for the crystal rims (0.3 ± 0.1 to 2.5 ± 0.6 k.y.) also fall within uncertainty of crystal rim and dacite matrix ages, implying short durations between last recharge, magma mixing, and eruption.
Since the Ba zonation of the sanidine resulted from thermal variations and not compositional mixing (except the outermost rims), the correlation of zonation patterns across all megacrysts (Fig. 2C) suggests they each shared a similar thermal history of heating and cooling cycles, where diffusion times relate to the timing of cycles. Accordingly, Figure 3D shows temperature versus time during crystal growth with a sequence of major heat pulses. The decreasing zone widths (Figs. 1 and 2) and decreasing partial diffusion times from core to rim (Dataset S5) suggest the heat pulses became more frequent with time. This increase in heat pulse frequency may have eventually reached a threshold, where the crystal-rich rhyodacite magma was mobilized to mix with incoming mafic magma and become eruptible. The outermost rims indicate a final heating and hybridization event followed by cooling and crystallization during ascent and eruption of the hybrid dacite containing the megacrysts. Below the critical threshold frequency, however, the recharge events acted only as heat pulses followed by cooling and megacrystic growth, but no eruption.
The cross-crystal correlation of Dt values from specific zone boundaries (Fig. 3A) indicates that a similar amount of diffusion occurred synchronously at each zone boundary in all crystals. This means that not only did all megacrysts experience a similar thermal history, but the temperatures of heating and regrowth were also similar. With a very high activation energy (455 KJ/mol; Cherniak, 2010), Ba diffusion in sanidine is extremely sensitive to temperature, and a small change in temperature would significantly affect the extent of diffusion, e.g., a 10 °C change will alter Dt by 60%–80% (Supplemental Material S6). We do not observe such large Dt variations. Therefore, based on the tight range and errors in boundary-specific Dt values, we estimate the temperatures experienced by different megacrysts to be uniform within 20 °C (Supplemental Material S6).
Such consistent temperatures would be unlikely in a large magma reservoir because all crystals would have to come from a single isotherm within the reservoir. Although latent heat buffering may also contribute to homogeneous temperatures at near-solidus levels within some parts of a reservoir (Huber et al., 2009), it cannot explain the consistency over such a broad temperature range of 720–827 °C. A simple model of a reservoir conductively heated by underplating while cooling to its surroundings shows that for different reservoir shapes and sizes (1–1.5 km in the largest dimension), even 75% recharge relative to the reservoir size can affect only a small (<200 m length scale) portion of its volume (Supplemental Material S7). Changing temperature consistently over a large magma body is nearly impossible. Eruption from a large reservoir would also involve crystals with largely different thermal histories (across different isotherms), which is not indicated. Thus, crystals within a small reservoir with uniform temperatures is the more likely scenario here. For a variety of sizes of the reservoir, we track temperature variations by at least 10 °C over a length scale of 150–200 m and, thus, put the reservoir size within 400 m in the largest dimension.
Apart from the general character of typical sawtooth Ba zonation and unusually large size, there is no correlation in the sequence of heat pulses or Dt values between older and younger dome dacites at TVC (Rout et al., 2021), including among domes from the same stage IV as the one studied here. This implies that the megacrysts actually were derived from different magma batches, each with a long (hundreds of thousands of years) thermal history of temperature cycles and representing a distinct small melt-rich pocket within a possibly larger mushy system. Magma batches became eruptible only when recharge became spatially focused, more frequent, more voluminous, or all of the above. Our results point to a delicate balance among recharge location, recharge rate, and crystallinity that control eruptibility and provide only a small window of eruptible conditions in a self-organized, steady-state system. This also resulted in the uniform composition and nature of erupted dacite magmas at TVC over 1.3 m.y.
K-Feldspar Megacrysts in Granites and Other Settings
The size, sawtooth Ba zonation, and concentric bands of inclusions in TVC sanidine megacrysts are surprisingly similar to K-feldspar megacrysts in many silicic intrusions (Vernon and Paterson, 2008; Moore and Sisson, 2008; Johnson and Glazner, 2010). This extends our interpretations to the origins of such megacrysts in granites.
The origin of intrusive K-feldspar megacrysts is explained by two contrasting models: (1) slow growth below the K-feldspar liquidus with magma mixing and crystal cycling between magmas (e.g., Vernon and Paterson, 2008; Chambers et al., 2020), or (2) late-stage subsolidus, in situ, fluid-dominated textural coarsening at temperatures as low as 400 °C (Johnson and Glazner, 2010).
Our results clearly suggest that megacrysts of such size and zonation pattern grow in relatively small but purely magmatic “hot” environments, i.e., well above the magma solidus, at shallow depths for extended periods of time (hundreds of thousands of years). However, the key is periodic heat pulses and temperature cycling around sanidine nucleation temperatures (720–770 °C) causing melting and regrowth. This cycle results in textural coarsening (Higgins, 1999, 2011), albeit at above-solidus conditions. The heat pulses come from distant hotter recharge events that supply heat, but no mass. The frequency of recharge is low enough not to induce mixing or eruption-triggering mobilization. After tens to hundreds of thousands of years of temperature cycling and decreasing recharge rates, these small magma batches may cool and amalgamate into K-feldspar–bearing intrusive bodies. Conversely, at increased recharge frequency, an eruption brings the magma to the surface above a large intrusive system. Taápaca represents such a “leaky” system. Only at the highest heat input do the large volumes of the mush become eruptible to produce the voluminous ignimbrites in the Central Andes (Weber et al., 2023; van Zalinge et al., 2022).
K-feldspar megacrysts are also observed in evolved rhyolites from ocean-island or alkaline intracontinental settings, but these examples are much smaller (sub-centimeter), show no consistent growth zonation, have poor Ba variations without significant resorption and inclusions (e.g., Baumann et al., 2022), and are largely characterized by low-contrast oscillatory zonation (e.g., Ginibre et al., 2004). We infer that, in these cases, the amplitude of temperature cycling, as well as growth and residence times, would be very different and not comparable to TVC sanidine megacrysts.
CONCLUSIONS
Taápaca sanidine megacrysts provide evidence for the existence of small, shallow hot magma batches feeding uniform dome eruptions from long-lived (1.3 m.y.) intrusive complexes. We show that a delicate balance between frequency and position of hotter mafic recharge, as well as the physical and thermal setting of the reservoir, keeps the system thermally active at above-solidus temperatures for hundreds of thousands of years and provides the necessary temperature cycling for megacrystic growth (see also discussion in Andersen et al., 2019). Recharge events are frequent, but only a critical level of recharge provides us with samples from the intrusive system through magma mixing, mobilization, and eruption.
Although Glazner (2021) argued against an upper-crustal above-solidus magma reservoir, citing the requirement of unrealistic heat and mass input (see also Annen, 2009), Taápaca megacrysts provide ample evidence for the opposite interpretation: sanidine growth above the melt solidus in shallow, but small (<400 m) reservoirs with frequent heat pulses and dissolution above the sanidine liquidus. Our results therefore provide the necessary constraints to build more realistic thermal models that include the specific physical, dimensional, and thermal parameters that will allow for long-lasting steady-state hot magmatic environments at upper-crustal levels to explain the observed zonation patterns.
ACKNOWLEDGMENTS
This study was partly funded by a German Research Society (Deutsche Forschungsgemeinschaft) grant (RO 5007/2-1) to S. Rout. We thank A. Kronz for his support during the electron microprobe analysis, and M. Blum-Oeste and E. Lohnert for the acquisition of element maps of the dacite matrix and megacryst 28-1. We are grateful to the editor and the anonymous reviewers for their valuable feedback and suggestions.