Young zircon ages and compositions at Oregon's Mount Hood as evidence of high-silica magma storage underneath an andesitic arc volcano Open Access

Zircon is a nearly ubiquitous phase in continental arc andesite, dacite, and rhyolite magmas. This is unexpected because eruption temperatures of most arc intermediate-to-silicic lavas are above the zircon-saturation temperatures (based on bulk composition) at which the mineral grows and is preserved. This begs the question: Why is zircon so commonly found (e.g., Bacon and Lowenstern, 2005; Claiborne et al., 2010a; Stelten and Cooper, 2012; Klemetti and Clynne, 2014)? The answer may lie in continental arc magmatism, where bodies of relatively cool and crystal-rich magma or mush may be stored at near-solidus conditions for much of their history (e.g., Bachmann and Bergantz, 2008; Cooper and Kent, 2014; Rubin et al., 2017; Jackson et al., 2018). In this situation, the interstitial liquid may be significantly more silica rich than the bulk composition and can be zircon saturated. During periods of magmatic recharge, the mush can be remobilized, taking its zircon cargo with it in the creation of hybrid intermediate magma (e.g., Wiebe et al., 2002; Kent et al., 2010).

Recent studies have shown that the zircon record at continental arc volcanoes can range up to 10⁵ yr (see review by Cooper, 2015). This observation supports the idea that zircon-saturated magma may persist for long periods relative to the time the volcano might be “restless” and ready to erupt (e.g., Murphy et al., 1998; Leonard et al., 2002; de Silva et al., 2008; amongst many others). The ages of these zircons can indicate the “lifespan” of mobilizable silicic mush (e.g., Klemetti et al., 2011; Allan et al., 2013; Klemetti and Clynne, 2014; Deering et al., 2016).

Oregon's Mount Hood is one of the most potentially hazardous volcanoes in the Cascade Range of North America (Ewert et al., 2018) with proximity to major cities and transportation routes. The current model for magmatism at Mount Hood is one of “recharge filtering” (Kent et al., 2010). This model envisions two end members: a mafic recharge magma and a silicic filter (or mush) stored in shallow crust (Scott and Gardner, 2017). These two magmas interact in the weeks to months prior to an eruption to produce hybrid crystal-richandesitic-to-dacitic magma that erupts as domes and lava flows.

Evidence for this model exists in the form of bimodal populations of amphibole, orthopyroxene, and plagioclase that formed at different temperatures and from different liquid compositions (Kent et al., 2010; Koleszar et al., 2012; Scott and Gardner, 2017; Kent and Koleszar, 2017), as well differing estimated radiometric ages for plagioclase from different populations of crystals (Eppich et al., 2012). The compositional and mineralogical similarities of magmas erupted over the ~450,000 yr lifetime of the current Mount Hood edifice suggest that this mechanism of eruption has persisted over time.

Despite the known volcanic hazard potential for Mount Hood, the longevity of this silicic filter and its thermal conditions are not known. U-series estimates suggest that the plagioclase in erupted lavas derived from the silicic mush has been stored in the crust for at least ~21,000 yr (Eppich et al., 2012; Cooper and Kent, 2014), but studies of zircon in other systems show magma residence can be much longer (e.g., Claiborne et al., 2010a; Klemetti and Clynne, 2014; Dechert et al., 2024).

Although Mount Hood persistently erupts intermediate lavas (Hildreth, 2007; Koleszar et al., 2012; Scott and Gardner, 2017), which are typically too hot and have bulk compositions that are too mafic to crystallize zircon, these lavas contain abundant zircon crystal cargo. In this study, we dated zircons found in the three most recent eruption periods at Mount Hood, spanning model ages of ~230 yr to >20 ka. Zircons were dated via the U-Th disequilibria radiometric dating technique. Young zircons are consistent with the presence and development of silicic mush at Mount Hood over the last 60,000 yr. The increased proportion of zircon since 20 ka suggests that the Mount Hood system may be cooling since the end of the Polallie Period ca. 12 ka, punctuated by brief, small rejuvenation events. Thermal models built from these observations show that the future activity at Mount Hood could be triggered by inputs of small volumes of new basaltic magma.

The stratovolcano of Mount Hood (3435 m) is part of the Cascade Range and is the northernmost volcano in Oregon (Fig. 1). It is located only 77 km due east of downtown Portland, Oregon, 57 km from Portland International Airport, and 35 km from I-84 and the Columbia River. The volcano has erupted ~ 50 km³ of lava and tephra over the past ~0.5 m.y. at an average rate of ~0.1–0.2 km³/k.y. (Hildreth, 2007; Scott and Gardner, 2017).

Volcanic activity in the vicinity of the modern Mount Hood may extend more than a million years (Hildreth, 2007; Scott and Gardner, 2017), but the current edifice likely began to form ca. 450 ka (Hildreth, 2007; Scott and Gardner, 2017). The current edifice overlaps two older systems: the Sandy Glacier volcano and another large andesitic volcano predecessor (Hildreth, 2007; Scott and Gardner, 2017). As with many arc stratovolcanoes, the volcano has erupted episodically, and the growth of Mount Hood has been characterized by periods of heightened activity that typically lasted decades to centuries interspersed with periods of quiescence and glacial erosion lasting 10,000 yr or more (Hildreth, 2007).

The most recent activity at Mount Hood was the Old Maid eruptive period. This was marked by low-silica dacite dome extrusion at Crater Rock near the modern summit, and block-and-ash flows and lahars that started in 1781 and may have lasted for up to several decades (Scott and Gardner, 2017; Sheppard et al., 2010). This period erupted only ~0.15 km³ of lava and tephra (Hildreth, 2007).

Prior to the Old Maid eruptive phase, the next most recent eruption was the Timberline eruptive period (Scott and Gardner, 2017). Like the Old Maid Period, the Timberline Period consisted of a series of low-silica dacite domes extruded in the Crater Rock location that produced block-and-ash flows and lahars, as well as some tephra fall. The Timberline phase was also associated with a significant sector collapse event that produced the smoothly sloping Timberline surface on the southern side of the volcano. This activity likely lasted a few decades as well and was about ten times more voluminous than the Old Maid Period, erupting ~ 1 km³ (Hildreth, 2007).

There is no record of activity on Mount Hood itself between ~12 ka and ~1.5 ka. However, ca. 7 ka, the Parkdale basaltic andesite lava flow erupted 12 km NNE of the main edifice (Scott and Gardner, 2017). This coulee spread 7 km from the vent below the lower slopes of the volcano and comprised 0.1 km³ of lava. Although it is debated whether this lava flow is directly related to the Mount Hood system or merely coincidental (Hildreth, 2007; Scott and Gardner, 2017), mineralogical and geochemical similarities between Parkdale samples and lavas from Mount Hood summit eruptions suggest these eruptive products were derived from the same broad magmatic system (e.g., Kent et al., 2010; Scott and Gardner, 2017).

One of the most vigorous periods of activity at Mount Hood was the Polallie eruptive period that spanned 30–12 ka (Scott and Gardner, 2017). This period was characterized by andesite and low-silica dacite domes that built much of the summit of the modern edifice, adding over 5 km³ of total erupted material (Hildreth, 2007). The Polallie phase generated pyroclastic flows and lahars that impacted all the flanks of the volcano (Scott and Gardner, 2017).

Overall, Mount Hood has erupted a monotonous series of andesite and low-silica dacite lava compositions throughout the lifetime of the current edifice, with measured compositions from summit eruptions ranging ~ 55–65 wt% SiO₂ (Scott and Gardner, 2017). Flank eruptions can be more mafic (≥55 wt% SiO₂), and Mount Hood lavas also show clear evidence of hybridization, including the presence of multiple crystal populations and crystal zoning, linear compositional trends, and an abundance of quenched magmatic inclusions. The latter have SiO₂ contents that are mainly 55–60 wt% SiO₂, although rare mafic cumulate-textured inclusions (<50 wt% SiO₂) also occur (Kent et al., 2010).

Mount Hood lavas are crystal rich, typically with 20–40 mod% crystals (Kent et al., 2010), dominated by plagioclase feldspar with subordinate mafic phases of amphibole, orthopyroxene > clinopyroxene, and occasional trace olivine. Zircon and apatite are also present as well as Fe-Ti oxides, such as ilmenite and magnetite.

The eruptive products at Mount Hood are remarkably consistent: crystal-rich andesite-dacite magmas. The model proposed by Kent et al. (2010) consists of a crystal-rich silicic mush of approximately rhyodacite bulk composition residing ~3–6 km below the volcano's summit. The mush zone acts as a filter, preventing more mafic magmas ascending from the lower crust from reaching the surface, and inducing hybridization to produce the crystal-rich andesitic and low-silica dacites that are erupted.

This “recharge filter” limits the compositions of magmas erupted because: (1) the mafic recharge magma is stalled physically and buoyantly by the less dense silicic filter, and (2) the silicic parent magma is too viscous to erupt on its own. This means that only hybridized magma can overcome both the density barrier and rheological lockup (Kent et al., 2010).

Plagioclase ²³⁰Th-²²⁶Ra and ²³⁸U-²³⁰Th geochronology of Old Maid and Timberline lavas suggests that multiple mineral populations have crystallized. One population of plagioclase, related to plagioclase grown during the mafic recharge events that immediately predated eruption, are within error of eruption ages, while a more texturally complex population has minimum and average ²³⁸U-²³⁰Th ages of 21,000 yr and 126,000 yr, respectively (Eppich et al., 2012; Cooper and Kent, 2014).

In the Kent et al. (2010) model, the silicic mush that acts as a filter at Mount Hood is thought to have been resident in the shallow crust for at least 21,000 yr (based on U-Th ages of erupted plagioclase from Timberline and Old Maid lavas; Eppich et al., 2012). Trace-element zoning in plagioclase derived from this mush also shows that these crystals were stored for long periods at temperatures (<~750 °C) where they are crystal-rich and rheologically locked. Eruptions occur when this mush is rapidly heated and remobilized by intruding basaltic or basaltic andesite magma (Scott and Gardner, 2017). The resulting hybrid magma erupts at ~ 900–1040 °C (Koleszar et al., 2012; Scott and Gardner, 2017).

Samples were collected by Kent and Klemetti during 2012–2016 (Fig. 1; Table S1¹). All samples were collected from blocks in block- and-ash pyroclastic flows from the Old Maid, Timberline, or Polallie eruptive periods. Blocks were assessed for prismatic cracking to indicate hot emplacement. Oxidized crusts were removed or avoided.

Each sample was hammer crushed to pea-sized fragments and then passed through a disc mill to reduce the material to mostly <500 μm. The crushed material was then sieved to >500 μm, 250–50 μm, and <50 μm fractions. Magnetic material (from crushing and milling as well as iron oxides) was removed with a hand magnet. All samples were run across a U-Tech RP-4 Shaker Table from which zircons were handpicked from the top of the table with a disposable pipettor and placed in a filter paper. These separates were washed with distilled water and ethanol and then passed through a small volume of methylene iodine to remove the light fraction. Finally, the heavy separate was rinsed with acetone in filter paper.

Mounts for the zircons were created at the Stanford University/U.S. Geological Survey (USGS) sensitive high-resolution ion microprobe (SHRIMP) ion microprobe (SUMAC) laboratory using methods from Coble et al. (2017). Each epoxy mount of zircon (unknowns and standards) was imaged by scanning electron microscope via cathodoluminescence (CL; Fig. 2). These CL images were used to assess placement of the spot for age and trace-element analyses, where melt or mineral inclusions were avoided. Overall, these zircons were between 50 µm and 200 µm long and 30–100 μm wide.

Isotopic and compositional analyses of Mount Hood zircons were performed via SHRIMP–reverse geometry (RG) at the Stanford University/USGS SUMAC laboratory in August 2014 and July–August 2018 (Tables S2 and S3). Age standards EBT (Early Bishop Tuff; age = 760 ka) and 91500 (Wiedenbeck et al., 1995; age = 1065 Ma) were analyzed repeatedly to monitor reproducibility and drift (Table S4).

In total, 70 zircons were analyzed for U-Th isotopes (²³⁸U, ²³²Th, ²³⁰Th) to determine U-Th disequilibria ages (Table S2). Of these 70 analyses, 32 were from samples MH2A and MH14–6 (Old Maid), 19 were from sample MH14–5 (Timberline), and 19 were from sample MH16–1 (Polallie). Eppich et al. (2012) reported whole-rock ²³⁸U-²³⁰Th activities from the Old Maid and Timberline eruptive periods of 1.05 for (²³⁸U/²³²Th) and 1.17 for (²³⁰Th/²³²Th). Allowing for 10% potential variation in the initial (²³⁸U/²³²Th) and (²³⁰Th/²³²Th) values based on data from Eppich et al. (2012), the range of potential initial activities is 0.95–1.16 for (²³⁸U/²³²Th) and 1.06–1.28 for (²³⁰Th/²³²Th). This yields an average change in calculated model age of up to 6% for (²³⁸U/²³²Th) and up to 19% for (²³⁰Th/²³²Th), respectively. Thus, the choice of initial (²³⁸U/²³²Th) and (²³⁰Th/²³²Th) values does not significantly affect the calculated ages or the interpretations herein.

Trace-element compositional analyses were performed on selected zircons (n = 95), coincident with the U-Th age spot where possible. Three of these zircons did not have corresponding U-Th ages, all from Old Maid samples. Of the trace-element analyses, 43 were from MH2A or MH14–5 (Old Maid), 30 were from MH14–6 (Timberline), and 22 were from MH16–1 (Polallie). All trace-element data are provided in Table S2. For trace-element compositional analyses, MAD-1 or MAD-559 (Coble et al., 2018) was used to calibrate the composition of unknown zircon (Table S3).

Ti-in-zircon temperatures were calculated using the method of Ferry and Watson (2007), assuming a(Si) of 1.0 and a(Ti) of 0.6 (Table S3). The assumption of quartz present in a silicic crystal mush supports an a(Si) value of 1.0. For zircon crystals included in other phases, it is assumed that the magma in which the major phase formed reflects the magma in which the zircons formed, so a(Si) = 1 is still applicable. In all cases, a variation in a(Si) of ~0.05 would change calculated temperatures by ~5 °C. For most calc-alkaline arc magmas, a(Ti) is between 0.5 and 0.8 (Ghiorso and Gualda, 2013).

We chose 0.6 to reflect the a(Ti) value from Mount St. Helens (Claiborne et al., 2010a) as the closest analog for Mount Hood. Errors in activity estimates of 0.2 would change calculated temperatures by 30 °C. However, these would shift all temperatures systematically and would not significantly affect the interpretations. Analytical errors on the ⁴⁸Ti measurements, relative to the MAD-559 zircon standard, were 3.8% (1σ), leading to average errors based on analysis on all temperatures of ~5 °C (1σ; Klemetti and Clynne, 2014).

Zircons imaged from the Old Maid, Timberline, and Polallie samples at Mount Hood are unremarkable in CL (Fig. 2). Some are weakly zoned with only minor changes in color from light to dark gray. Unlike zircon found in many other volcanic systems in the Cascades (e.g., Mount St. Helens in Claiborne et al., 2010a; Lassen Peak/Chaos Crags in Klemetti and Clynne, 2014; Crater Lake in Bacon and Lowenstern, 2005; Devil's Hills on South Sister in Stelten and Cooper, 2012), the weak zoning appears as few, wider bands that gradationally transition into the next band. Rare zircons are weakly sector-zoned as well. There is also a slight preponderance of light bands on the rims of the crystals, although some of this brightness may be due to the topography of the zircon in the mount.

Mount Hood zircons (n = 70; Fig. 3) presented an array of values. Isochrons for crystals created using the initial whole-rock (²³⁰Th/²³²Th) and (²³⁸U/²³²Th) values from Eppich et al. (2012) have slopes that range from apparent negative values (n = 25) to 1 (secular equilibrium; n = 7). Only seven crystals gave negative slopes that are outside of 1σ error of a zero slope, and none gave negative slopes at 2σ error (Table S2).

Zircons are also relatively low in U (21–819 ppm; average = 96 ppm, median = 76 ppm), so for very young zircons (<2000 yr), very little ²³⁰Th will be present (Fig. 3). The isochron slope for individual, young crystals could easily plot as negative due to the low ²³⁰Th signal versus the background measurement and possible variation in the initial whole-rock activity ratio. All zircons with negative apparent slopes (except one) had U content <150 ppm (Table S2).

When these isochrons are converted into ages, ~47% of the crystals (n = 33) are less than 100 k.y. old (Fig. 3). We have assigned the respective “eruption age” to zircons (n = 25) with negative apparent ages (Table S2). As mentioned in the preceding paragraph, these negative ages (based on negative slopes) are likely statistical artifacts of the low ²³⁰Th concentrations in the zircons. Many times, the recorded counts per second (cps) for each zircon were close to the ~0.04–1 cps noise level of the detectors. Given this, we interpret that these zircons are likely very young and crystallized within error of their respective eruption age. For zircons with negative apparent ages, if 2σ errors on the isochron data are propagated into the age calculation, then only six zircons remain with slightly negative ages (Table S2).

Both the Old Maid and Timberline samples showed peaks at ca. 1.2–0.4 ka and ca. 12–11 ka (Figs. 4A and 4B) with long tails to >100 ka. The Polallie sample had peaks at ca. 47 ka, 65 ka, and 100 ka (Fig. 4C). When considering the error on the U-Th disequilibria model ages, these peaks may overlap. However, across all samples, there were clear peaks at <12 ka, ca. 60–40 ka, and ca. 100 ka, with sparse samples extending from ca. 100 ka to secular equilibrium (ca. 350 ka; Figs. 3 and 4).

Hafnium can be used as a proxy for the extent of fractionation in magma, where high-Hf zircons represent high fractionation and vice versa (Claiborne et al., 2010a, 2010b; Barth et al., 2012; Klemetti and Clynne, 2014). The overall Hf concentrations for Mount Hood zircons spanned ~8000–12,500 ppm (Fig. 5), and a majority of zircon (within lower and upper quartiles) Hf concentrations were between 9200 ppm and 11,000 ppm, with a median of ~9700 ppm.

The Eu anomaly recorded in zircon captures the extent to which plagioclase has crystallized in the host magma prior to zircon crystallization (Claiborne et al., 2010b; Klemetti et al., 2011; Klemetti and Clynne, 2014). Values closer to 1 indicate that less plagioclase was fractionated. Eu/Eu* values in Mount Hood zircons are from 0.30 to 0.65 (Fig. 5), and a majority (within lower and upper quartiles) of Eu/Eu* values range between 0.36 and 0.47, with a median value of ~ 0.41.

Ti-in-zircon temperatures ranged from 679 °C to 868 °C, although most zircons (within the upper and lower quartiles) recorded temperatures between 726 °C and 766 °C (Figs. 5 and 6). Ti-in-zircon temperatures are highly dependent on the activity of Si and Ti in the host magma. We detailed our chosen values for each of these in the Methods sections, but even if the absolute values derived by the Boehnke et al. (2013) method of temperature calculations might be slightly inaccurate, the errors in temperature are likely systematic. This means that comparisons of relative temperature values are robust, unless the activities of Ti and Si vary over time.

Comparatively (Fig. 4), the Polallie period zircons showed the highest median Ti-in-zircon temperature of ~760 °C, followed by the Timberline period zircon at ~740 °C and the Old Maid period zircons at ~730 °C, suggesting an overall cooling of the system across these three periods.

The relationship between Yb/Gd and Th/U follows the extent of basaltic input, where decreasing Yb/Gd values with increasing Th/U values suggest an increased degree of basaltic input to the magmatic system (Barth et al., 2012; Klemetti and Clynne, 2014; Klemetti et al., 2014). In the Mount Hood zircons, the overall span of Yb/Gd is ~6–18, with a majority (within the lower and upper quartiles) of values between 8.4 and 14.1, and a median of 9.9. The range for Th/U is 0.4–0.8, with a majority of values between 0.5 and 0.7, and a median of 0.64. When Hf (ppm) is plotted versus the ratio of Th/U to Yb/Gd, higher values of Hf are correlated with lower values of (Th/U)/(Yb/Gd) (Fig. 6).

The fundamental observation from the U-Th zircon geochronology for the Mount Hood magmas is that they are remarkably young compared to zircons from other Cascade volcanic systems such as Mount St. Helens (Claiborne et al., 2010a), South Sister (Stelten and Cooper, 2012), and the Lassen volcanic center (Klemetti and Clynne, 2014), where most zircons are tens to hundreds of thousands of years older than the eruption age.

Although there are a handful of zircons at Mount Hood that date back to >100 ka, suggesting a long-lived presence of zircon-saturated silicic magma under Mount Hood, most of the zircons analyzed crystallized within 20,000 yr of the eruption age of their host lava. Some crystals may have formed within a few centuries to millennia prior to eruption. This recent crystallization is also supported by the simple textures and zoning seen in most of the Mount Hood zircons, which are not consistent with multistage growth of zircon. Although this might seem to be in contradiction with the bimodal crystallization of many major phases (Kent et al., 2010; Koleszar et al., 2012; Scott and Gardner, 2017; Kent and Koleszar, 2017), these zircons likely all formed from a high-silica liquid/mush separate from the mush of the major phase of growth.

When comparing the ages of crystals derived from Old Maid (0.2 ka), Timberline (1.5 ka), and Polallie (30–12 ka) eruptive period samples, two patterns become evident. First, both Old Maid and Timberline samples have zircon populations that are within error of the eruption age or within a few thousand years of their respective eruption ages (Fig. 3). Second, zircons from the Polallie age sample tend to be older, predating the Polallie period eruptive activity by 10–70 k.y.

However, for Old Maid and Timberline samples, most zircons formed during or since the end of the Polallie eruptive period (30–12 ka; Fig. 3). There are very few examples of such recent zircon growth (both relative to present and relative to eruption age) at intermediate arc volcanoes. In fact, many times, there is a lack of near- or at-eruption-age zircon, and this has been interpreted to indicate an extended period in zircon-undersaturated magma, thus preventing young zircon crystallization (Claiborne et al., 2010a; Klemetti and Clynne, 2014).

This implies rapid zircon nucleation and growth at Mount Hood close to the time of eruption. One explanation for this might be that recently cooled silicic mush that reached zircon saturation was also the first to be rejuvenated because it would take less thermal energy to heat that mush to mobilization and eruption. This would preferentially sample recently formed zircons that have not been repeatedly recycled, potentially explaining the overall mundane CL zoning textures observed (Fig. 2).

In zircon studies from silicic and intermediate arc magmas, the outer surfaces of zircon are thought to be close to or overlap with the eruption age of their host lava (e.g., Stelten and Cooper, 2012; Tierney et al., 2016; Burgess et al., 2021). However, studies that show zircon interiors that are within error of the eruption age, as we observe at Mount Hood, are rarer. Zircons from Dominica in the Lesser Antilles show crystallization at or near the eruption age of the major ignimbrites (Frey et al., 2018). Very large explosive rhyolitic eruptions from the Okataina volcanic center in New Zealand (Brown and Fletcher, 1999; Charlier and Wilson, 2010; Rubin et al., 2016) and the Long Valley caldera in California (Crowley et al., 2007) also feature zircons growing within error of the eruption age.

The compositional variety of zircon from these Mount Hood samples is relatively limited (Figs. 5 and 7). This would suggest that the conditions in which zircon crystallized and was stored (i.e., silicic mush or filter) at Mount Hood have been relatively unchanged over the past ~100 k.y. All three eruptive periods broadly overlap in temperature (~650–850 °C), Hf (~8000–12,000 ppm), and Eu/Eu* (~0.30–0.50). However, the sharp increase in frequency of zircon since 20 ka supports the idea that the system was consistently in zircon-saturation conditions during the Timberline and Old Maid Periods.

There are some trace-element variations. We observed a decrease in Hf with a decrease in the ratio of (Th/U)/(Yb/Gd) across all samples. However, the Polallie samples are offset to lower Hf and higher (Th/U)/(Yb/Gd), while Old Maid and Timberline are the converse. This would support the idea that Polallie samples were derived from magma that was less fractionated and was experiencing more basaltic input, while the Old Maid and Timberline products came from more fractionated magma that had less basaltic input. This relationship, combined with the zircon age spectra, indicates a model where much of the silicic mush being sampled during the Old Maid and Timberline Periods was Polallie-era magma that was continuing to cool and crystallize but was occasionally perturbed by small injections of new basalt (Fig. 7).

Studies by Kent et al. (2010) and Koleszar (2011) have constrained the thermal conditions for the compositional end members and hybrid magma at Mount Hood. The silicic end member was likely stored at less than ~750 °C and then heated briefly by intrusion of >950 °C basaltic magma to a final silicic mush temperature of ~ 850–900 °C. The resulting hybrid magma erupted at ~900–1040 °C.

Zircon-saturation temperatures are <700 °C for all erupted magma (Table 1), so zircon would have actively dissolved in the hybrid erupted magmas. The persistence of zircon requires that episodes of recharge and rejuvenation were brief, as zircon with an average diameter of 50 μm should dissolve in magma with this state of zircon undersaturation within years to centuries (Klemetti and Clynne, 2014). This brevity is also consistent with the time scales of weeks to months estimated for recharge events leading to eruption (Kent et al., 2010).

The average composition of lava erupted during the Old Maid, Timberline, and Polallie eruptive periods was andesitic to low-silica dacite (Table 1). These liquids would have been undersaturated with regard to zircon based on the calculated zircon-saturation temperatures (687–708 °C; Table 1) and Fe-Ti–oxide temperatures (Table 1; Koleszar, 2011). Ti-in-zircon temperatures for the zircons (Fig. 4) are lower than the Fe-Ti–oxide temperatures, supporting a model where the zircons were crystallizing from a melt that was cooler and very likely more silicic in order to crystallize and preserve zircons until eruption.

The calculated mafic end member for Mount Hood (Table 1; Kent et al., 2010; Koleszar, 2011) produces a zircon-saturation temperature range of 605–624 °C. The hybrid magma (Table 1) produced by interaction of the mafic and silicic end members to produce the intermediate lava that erupts at Mount Hood would have zircon-saturation temperatures of 689–708 °C, which overlap within error with the average Ti-in-zircon temperatures (Fig. 4). These are both well below the eruption temperatures calculated by Koleszar (2011).

The median of Ti-in-zircon crystallization temperatures for Mount Hood zircons is 745 ± 40 °C (Table S3). Using the silicic end member calculated by Kent et al. (2010) for major elements and by Koleszar (2011) for trace elements (Table 1), the range of calculated zircon-saturation temperatures (using methods from Boehnke et al., 2013) for this end member at Mount Hood is 695–740 °C. These two ranges overlap and are likely to represent the conditions in the silicic end member, which is 70 wt% SiO₂. Although this composition has never erupted at Mount Hood, it is where zircon is crystallizing during intervals between eruptions.

At Lassen Peak, the time scales of rejuvenation to preserve zircon are in the order of ≤10²–10³ yr, with the eruption effectively removing heat from the mush (Klemetti and Clynne, 2014). Diffusion models for plagioclase crystals at Mount Hood suggest remobilization in months to weeks (Kent et al., 2010). Using zircon dissolution models from Harrison and Watson (1983), zircon with an average diameter of ~100 μm would dissolve in years to centuries in an average hybrid andesite magma at 900–1050 °C from Mount Hood (Koleszar et al., 2012).

We interpret that the silicic mush required to form zircon at Mount Hood is also the silicic filter described by Kent et al. (2010), which interacts with basaltic injections to produce the hybridized andesite and low-silica dacites that erupt. The presence of some older (>100 ka) zircons as well as U-Th-Ra data derived from plagioclase feldspar (Eppich et al., 2012) suggest that mush in some form has been present under Mount Hood for an extended period, as suggested for other Cascade volcanoes (Claiborne et al., 2010a; Stelten and Cooper, 2012; Klemetti and Clynne, 2014). However, our zircon ages also suggest a significant proportion of the active silicic mush may have formed within the last 20,000 yr and has been especially productive for zircon formation in the past 10,000 yr.

The overall zircon age distribution at Mount Hood is distinct from other Cascade volcanoes as well, as noted in the following subsections.

At Mount St. Helens, rare zircons have been found with ages within error of their respective host sample eruption age (Flanagan, 2009; Claiborne et al., 2010a). Most zircons date from 10⁴–10⁵ yr prior to eruption, supporting a model of long-term recycling of zircon from a zircon-saturated crystal mush.

Much like Mount St. Helens, the zircons sampled from the three most recent eruptions of the Lassen volcanic center reveal a long history of crystal recycling to over 10⁴–10⁵ yr (Klemetti and Clynne, 2014). In these samples, the youngest zircon surfaces dated predate the eruption age by at least 15 k.y., suggesting little- to-no zircon growth close to eruption or no sampling of young zircons that may have grown in the silicic mush under the Lassen volcanic center.

Zircons from the ca. 2.3–2 ka Rock Mesa and the Devils Hills rhyolite at South Sister in Oregon dominantly record ages of 80–20 ka, with a minority of crystals recording older (up to secular equilibrium) ages and one recording the eruption age (Stelten and Cooper, 2012). Low Ti-in-zircon temperatures coupled with the compositions of the Rock Mesa and Devils Hills rhyolite that are zircon undersaturated suggest that the zircons were mostly derived from a longer-lived (10⁴–10⁵ yr) crystal mush reservoir.

This sets Mount Hood apart for its preponderance of zircons that have ages within error of the eruption age of the host lava (Old Maid or Timberline). Additionally, the lack of larger proportions of zircon from >100,000 yr at Mount Hood (Fig. 3) is in stark contrast to the long zircon legacy found at Mount St. Helens (Claiborne et al., 2010a), the Lassen volcanic center (Klemetti and Clynne, 2014), and South Sister (Stelten and Cooper, 2012).

However, at all volcanoes, the zircons appear to be forming dominantly during periods of eruptive quiescence, although not exclusively so (Claiborne et al., 2010a; Klemetti and Clynne, 2014). We cannot resolve exactly why older zircons are not observed, but it is either because the older zircon record was destroyed by heating or frozen by cooling, or possibly never reached conditions favorable for crystallization (i.e., zircon saturation).

This suggests a number of possible models for the preponderance of young zircon at Mount Hood: (1) Few zircons are preserved from the >100 k.y. history at the volcano due to dissolution during remobilization, (2) the silicic mush at Mount Hood was not zircon-saturated in the early history of the modern Mount Hood system, (3) the >100 ka silicic mush at Mount Hood cannot be remobilized because it has cooled well below the solidus, or (4) the silicic mush at Mount Hood was not present at the same depth or locations as the Old Maid and/or Timberline magma sources.

As Kent et al. (2010) and Kent (2014) proposed, the erupted lava at Mount Hood is formed by ~50:50 mixing of a basaltic input end member with a silicic magma end member. This mix produces a relatively monotonous andesite to low-silica dacite lava composition. These rejuvenation events likely last only weeks to years after the initial intrusion of basaltic magma, which many times leads to eruption.

These rejuvenation events of the silicic mush during basaltic intrusion at Mount Hood liberate zircons from the mush, which, due to their small size, tend to travel with the liquid (Claiborne et al., 2010a). The eruptive activity at Mount Hood has been much lower since the end of the Polallie eruptive period at 12 ka. Very small volumes of magma were produced during the Timberline and Old Maid eruptive periods (≤1 km³ each; Thouret, 2005), while the Polallie was moderately larger over much longer duration (4–8 km³ over 18,000 yr; Thouret, 2005). This supports the interpretation of smaller basaltic injections, which in turn may remobilize zircons without complete resorption due to their short durations (weeks to years; Cooper and Kent, 2014).

The presence of rare older zircons with the same composition as more recent Mount Hood zircons likely eliminates the hypothesis that zircon was not crystallizing. Instead, the zircon record was either erased, not sampled, or trapped (or some of each). This supports a model where there was a limited volume within the magmatic system that was rejuvenated during the Old Maid and Timberline eruptive periods. This lack of evidence for compositional input from the mafic magma is reinforced by observations of Fe zoning in plagioclase from these rocks (Kent et al., 2010).

Why has zircon crystallization apparently increased during the last ~20,000 yr at Mount Hood? This could reflect a change in the thermal regime in the volcano's magmatic system. The lower level of activity since the end of the major Polallie eruptive phase at ca. 12 ka may allow the silicic mush to return to zircon saturation more rapidly after rejuvenation, promoting more zircon growth.

Alternatively, less of the overall silicic mush might be heated above zircon saturation with each basaltic injection, leaving more volume of silicic magma to continue to crystallize zircon. This latter hypothesis is also proposed for the Lassen volcanic center. We propose an overall model (Fig. 7) at Mount Hood that allows for zircon crystallization with and mobilization from a silicic end-member magma repeatedly under the volcano. Zircons formed in this silicic magma get repeatedly released and recycled during short-durationrejuvenation events that move the silicic magma out of zircon saturation. Crystal age evidence from both zircon and plagioclase (Eppich et al., 2012) supports this model of long-lived crystal recycling followed by continued crystal growth.

The median Ti-in-zircon temperature of ~ 745 ± 40 °C is the maximum temperature of storage that will preserve zircon. The injection of ~1300 °C basaltic magma (Bacon et al., 1997) would raise the temperature of the silica reservoir to above the temperature of zircon saturation. The ensuing hybrid magma would be produced at temperatures in the range of 900–1040 °C (e.g., Koleszar et al., 2012) and thus, would likely actively dissolve zircon in contact with the melt.

Short time scales (weeks to years; Cooper and Kent, 2014) of remobilization combined with the time needed for zircon dissolution would prevent the wholesale dissolution of zircon within the crystal mush. Additionally, the brief duration of remobilization heating would promote the growth of new zircon during the decreasing temperature path after eruption as the mush would cool quickly to zircon-saturated conditions. However, many of these zircons are so youthful (Fig. 3) that they have not experienced sufficient recycling to produce common resorption or regrowth textures, unlike the more complex textures at systems like Mount St. Helens (Claiborne et al., 2010a) and the Lassen volcanic center (Klemetti and Clynne, 2014).

The zircon model ages presented here suggest that a zircon-saturated silicic reservoir at Mount Hood has been present since at least 50 ka but has grown rapidly since 20 ka. What this reveals is that for Mount Hood–type volcanoes that produce relatively limited intermediate eruptive products, a significant reservoir of higher-silicic magma persists at or above the solidus for long periods of time (≥10⁴ yr). This repeated production of a hybrid intermediate magma might be controlled by the balance between the volume of the silicic reservoir and basaltic injection.

Higher rates of basaltic injection, which would add more heat and volatiles, could lead to the dome formation and pyroclastic flows sourced from the higher-silica reservoir magma (Pallister et al., 1992; Leonard et al., 2002; Ruprecht and Bachmann, 2010). However, without increased rates of basaltic injection, the volcano will likely continue to erupt this hybrid intermediate magma with less propensity for explosive activity.

This could also suggest that the shape of the Mount Hood magmatic system is much more centralized, where hybridization is favored over mingling because of the limited space. Magma is channeled through a narrow area (long, vertical) where new basalt magma interacts with the silicic mush to produce the hybrid andesite at Mount Hood, similar to that inferred at Mount St. Helens (Dzurisin, 2018). In this way, the silicic “cap” could be perceived as more of a silicic “coating” that is heated, remobilized, and hybridized during ascent.

This is different than a magmatic system with a larger footprint like the Lassen volcanic center, where interactions at different regions laterally within the system create more variable erupted compositions. The long and narrow orientation of the reservoir could be favored in many Mount Hood–type volcanoes in arcs that erupt monotonous intermediate lava and tephra.

The ²³⁸U-²³⁰Th model ages from zircons sampled from the last three eruptive periods at Mount Hood provide evidence for young zircon crystallization at an intermediate arc volcano. Unlike many other Cascade (and other) volcanoes, the zircon population is dominantly less than 20,000 yr old, with a lack of the multitude of older (>100 ka) zircons that has become a hallmark of crystal recycling.

This difference in zircon ages reflects the differences in the silicic magma stored underneath these volcanic systems. At Mount Hood, the silicic end-member magma that prevents the eruption of the intruding basaltic magma either lacks the long heritage of zircon growth, or that record is locked and unable to be remobilized. Our proposed model for Mount Hood reflects the changing conditions when the most eruptively active Polallie Period (30–12 ka) ended and transitioned into the lower eruptive productivity of the Old Maid (0.2 ka) and Timberline (~1.5 ka) Periods. The sharp increase in zircon crystallization combined with an overall similar composition of zircons across all three periods suggest that the silicic mush for the later eruptions formed from cooling Polallie-era magma. The occasional rejuvenation manifested in the Old Maid and Timberline eruptions stemmed from small, new basaltic intrusions. However, the overall crystallization conditions for zircon in the Polallie eruption products imply a less fractionated silicic cap during that period.

The ability for a volcano like Mount Hood that erupts andesite to form abundant zircon is evidence for a zircon-stable long-term silicic magma reservoir beneath the volcano. This magma is likely ~70 wt% silica, yet silica has not been recorded in the eruptive products over the past 500 k.y. This balance between basaltic input and silicic filter to produce andesite is a hallmark of Mount Hood and its style of eruption.

Whether this balance is a product of a constant basaltic flux, a specific volume of silicic magma, or both is unclear. However, these zircon data show that ephemeral mafic-silicic mixing events may dominate the eruptive record at Mount Hood, but during much of the volcano's history, zircon has been crystallizing in a much more silicic regime.

We would like to acknowledge the following for their help in field or laboratory work on this project: Elisabeth Bertolett (Denison University, University of Canterbury, Christchurch, New Zealand) and Jeff Sullivan (Tampa Bay Rays). Funding for this project came from the Denison University Research Fund to E.W. Klemetti and National Science Foundation grant EAR-1763639 to A.J.R. Kent. We would also like to thank Holli Frey (Union College, Schenectady, New York, USA) and an anonymous reviewer for their comments, which greatly enhanced this manuscript.