Phengite-Based Chronology of K- and Ba-Rich Fluid Flow in Two Paleosubduction Zones

See allHide authors and affiliations

Science  03 Jan 2003:
Vol. 299, Issue 5603, pp. 92-95
DOI: 10.1126/science.1076977


Subduction recycles aqueous fluids from slab and sediment to the mantle. Subduction zones are long-lived, but time scales for fluid-rock interaction within subduction complexes are uncertain. Large-ion lithophile elements (potassium and barium) were added to eclogite (subducted basalt) during high pressure/temperature metamorphism via phengite crystallization from subduction zone fluids. Phengite grains from eclogite blocks and their metasomatic selvages yielded 40Ar/39Ar ages across grains and between samples that indicate 25 and 60 million years of fluid-rock interaction in the Samana Complex, Dominican Republic, and the Franciscan Complex, California, respectively.

Phengite, the principal host of large-ion lithophile elements (LILEs; here, K and Ba) in subduction zone metabasalts, carries these elements to depths of more than 180 km (1–3). Phengite dehydration at high pressure and temperature (P/T) conditions produces LILE- and H2O-rich fluids, which can transfer these components to the overlying mantle wedge and through shallower levels of the subduction complex (2,4). Large (0.2 to 4 mm) phengite grains in eclogite blocks and zoned selvages (rinds) around the blocks from two high-P/T metamorphic terrains (the Samana Metamorphic Complex in the Dominican Republic and the Franciscan Complex in California) contain 0.2 to 1.6 weight percent (wt %) BaO (table S1) (5). The Ba zoning patterns of phengite grains suggest that they formed from LILE-rich fluid-rock interaction during subduction zone metamorphism (3, 6). To investigate whether Ba zoning reflects time-resolvable fluid-rock events, we obtained 40Ar/39Ar ages for these grains.

The Samana Metamorphic Complex (Fig. 1) contains eclogite blocks produced during subduction of the North American plate beneath the Caribbean plate (supporting online text) (7, 8). Glaucophane Sm/Nd ages suggest eclogite facies metamorphism at 84 ± 22 million years ago (Ma) (9), and phengite K-Ar ages of 38 ± 2 Ma record retrogression and uplift due to slip along the Septentrional fault zone, the present North American– Caribbean plate boundary (7, 8, 10). Two blocks from Punta Balandra (SS84-24A and SS85-27E), their transition zones and inner rinds (SS84-24B, SS84-24C, and SS85-27B1), and their actinolite-rich outer rinds (SS84-24D and SS85-27B2) yielded phengite grains (Fig. 2) (3).

Figure 1

Geologic maps for (A) the Samana Peninsula (7) and (B) the Franciscan Complex (14, 17). Samana (SS) samples are from the Punta Balandra unit (3); Franciscan samples are from Ring Mountain, Tiburon Peninsula (T-90), and Mount Hamilton, Diablo Range (MH-90 and GL-16) (3).

Figure 2

Schematic, not-to-scale diagrams (19) of Samana (upper) and Franciscan (middle andlower) blocks, showing rock types, and average40Ar/39Ar phengite ages (±1σ), with MSWDs in parentheses (ages for individual grains are shown in Tables 1 to 3). Arrows point toward younger average ages.

The Franciscan Complex (Fig. 1) contains garnet blueschist, garnet amphibolite, and eclogite blocks (Fig. 2) (supporting online text) (11). It formed during east-dipping Mesozoic subduction beneath the western margin of the North American plate and is cut by the Miocene-initiated San Andreas Fault system. Jurassic to early Tertiary K-Ar ages for Franciscan whole rocks, glaucophane, hornblende, and white mica (12–17) may reflect eclogite or amphibolite facies metamorphism at 163 to 158 Ma, cooling to below phengite's Ar closure T between 159 and 139 Ma, and continuing blueschist P/T conditions from ∼146 to 80 Ma (17).

Franciscan Complex samples analyzed here are from Ring Mountain in the Tiburon Peninsula (12, 18) and Mount Hamilton in the Diablo Range (3, 13,19–21). Tiburon samples include garnet amphibolite (T-90-2AH) and eclogite blocks (T-90-3AG and T-90-3AB), actinolite-rich rind (T-90-2B), and cross-cutting phengite and a chlorite vein (T-90-2V). Mount Hamilton samples include an eclogite block (MH-90-1AB), a garnet amphibolite layer within the block (MH-90-1AA), hornblendite relicts of a high-T rind (MH-90-09) and vein (MH-90-11C), and the block's low-Ttransition zone (GL-16-01) and actinolite rind (GL-16-04) (Fig. 2).

We imaged with backscattered electrons (BSEs) 38 0.5-to-4-mm phengite grains before and after Ar analysis (5). Electron microprobe analyses and x-ray maps of other phengite grains from the samples linked brighter regions in the BSE images to higher BaO contents. Each grain to be analyzed for Ar was then placed in an individual glass pod (fig. S1) and irradiated. An Ar ion laser with a spot size ∼30 to 50 μm in diameter extracted the samples, which were analyzed by the 40Ar/39Ar method. Grains were then BSE-imaged again to ascertain the location and degree of alteration around lased spots. Four lased Samana grains (SS84-24D, grain 1; SS85-27B2, grains 1 and 2; and SS84-24C, grain 2) were foil-wrapped and step-heated under vacuum from 500° to 1350°C. (See tables S2 to S9 for the full data set, including Ar isotope ratios and irradiation and correction factors. Ar ages and isotope ratios are quoted ±1σ.)

The 40Ar/39Ar ages of Samana phengite grains range from Eocene (48.9 ± 3.7 Ma, SS85-27B1) to late Oligocene (24.7 ± 3.7 Ma, SS85-27B2) (Table 1). In each block-to-rind suite, average phengite ages for the inner rind are oldest and those from the eclogite block are youngest (Table 1and Fig. 2). Average ages decrease from the inner to the outer rind by ∼5 to 8 million years (My), and from the inner rind to the eclogite block by ∼12 to 13 My. Spot ages for individual phengite grains are rarely consistent with a single population, except for outer rind SS84-24D, which yields an age of 38.5 ± 2.9 Ma, with a mean square weighted deviation (MSWD) of 1. Grains from the transition zones and outer rinds show the largest MSWD. Age contours overlaid on BSE images show that the grains lack older-core to younger-rim patterns (fig. S2).

Table 1

Samana 40Ar/39Ar phengite ages, with ±1σ uncertainties. The 48 Samana analyses together average 37.8 ± 2.6 My, with a MSWD of 13. See Fig. 2 for average ages and MSWD for each sample. For each sample, numbers in parentheses indicate the grain and spot number, respectively.

View this table:

To examine age heterogeneities and estimate the extent of excess 40Ar contamination, individual lased grains from samples with a large MSWD (SS84-24C and SS85-27B2) and the single-population sample SS84-24D were step-heated (Fig. 3). Grains 1 and 2 from SS85-27B2 yielded 3- to 6-My (±1σ) differences in total gas ages (TGAs), calculated by combining results from each step in proportion to the amount of gas present. Differences were not due to excess 40Ar, because correlation isochrons for these grains and other step-heated samples intersected36Ar/40Ar values at an atmospheric value of 295.5 ± 1.5. Only grain 2 from SS85-27B2 displayed an average lased age (43.1 ± 1.2 Ma) that overlapped its step-heated TGA (41.4 ± 0.8 Ma). Although the other step-heated phengite grains yield average lased ages younger than their step-heated TGAs, results for individual spots resemble the TGA (Table 1 and Fig. 3).

Figure 3

(Left) Step-heated Ar age spectra and (right) correlation diagrams of Samana phengite from (top) SS85-27B2, grains 1 and 2; (middle) SS84-24C, grain 1; and (bottom) SS84-24D, grain 1. TGAs (±1σ) are shown in parentheses.

Average lased ages of phengite grains in Tiburon garnet amphibolite block T-90-2AH, from the Franciscan Complex, are 3 to 5 My (±1σ) older than those of its low-T rind (T-90-2B) (Table 2). Phengite grains from the vein average 155 ± 2 Ma (T-90-2V), which is indistinguishable from the block's average phengite age of 153 ± 2 Ma (Fig. 2). Ages of 156 ± 6 Ma and 152 ± 2 Ma for phengite grains from interlayered eclogite and garnet blueschist block T-90-3 resemble those from garnet amphibolite block T-90-2AH. Phengite grains from the T-90-2 rind and vein have the largest MSWDs of Tiburon samples (Fig. 2). The inner rind displays the oldest and youngest phengite spot ages (T-90-2B, 160.6 ± 3.3 Ma and 141.0 ± 1.7 Ma); phengite crystallization there both preceded and outlasted that within its host block T-90-2AH. As in the Samana samples, approximate age contours for Tiburon single grains lack older-core to younger-rim patterns (fig. S2).

Table 2

Tiburon Peninsula Franciscan40Ar/39Ar phengite ages, with ±1σ uncertainties. The 37 Tiburon analyses together average 153 ± 3 My with a MSWD of 5. See Fig. 2 for average ages and MSWD for each sample. For each sample, numbers in parentheses indicate the grain and spot number, respectively.

View this table:

The age ranges and large MSWDs for Mount Hamilton phengite grains (Table 3 and Fig. 2) suggest that this block was probably exposed to complex fluid-rock interaction for longer than those from Tiburon or Samana. The oldest Mount Hamilton grains are from the high-T vein (MH-90-11C, 148.7 ± 1.3 Ma) and low-T rind (GL-16-04, 148.1 ± 3.6 Ma), whereas the youngest grains are from the low-T rind (GL-16-04, 114.1 ± 8.2 Ma) and high-T rind (MH-90-09, 99.8 ± 10.2 Ma). Some spot ages from the rind, transition zone, eclogite block, and garnet amphibolite overlap at ∼144 Ma (Table 3). This age also overlaps with the Tiburon low-T rind and temporally connects the two localities.

Table 3

Mount Hamilton Franciscan40Ar/39Ar phengite ages, with ±1σ uncertainties. The 32 Mount Hamilton samples together average 139 ± 6 My with a MSWD of 17. See Fig. 2 for average ages and MSWD for each sample. For each sample, numbers in parentheses indicate the grain and spot number, respectively.

View this table:

Ar results for Samana phengite grains (Table 1 and Fig. 3) confirm what their Ba-rich compositions and patchy zoning suggest: These eclogite blocks were affected by multiple episodes of LILE metasomatism during a long (∼25 My) history of fluid-rock interaction. During Eocene subduction of the North American plate beneath the Greater Antilles, LILE-bearing fluids interacted with the eclogite blocks, which were then encased in serpentinizing peridotite (3) to form actinolite- and phengite-rich rinds. During the Oligocene, both blocks and rinds interacted with other pulses of LILE-rich fluid, which enlarged rind exteriors. Phengite did not form in the block, which was presumably protected from fluid access by the growing rind, until the late Oligocene. Age gradients modeled for single phengite grains lack older-core–to–younger-rim patterns (fig. S2). These relationships among closely spaced and genetically related rock types testify to ∼25 My of fluid-rock interaction.

The Franciscan data suggest that during subduction of the Farallon plate under the western margin of the North American plate, LILE-rich fluids began to form phengite within the otherwise mid-ocean ridge basalt–like (3) Tiburon blocks at 154 ± 4 Ma and within the Mount Hamilton eclogite block 9 to 18 My (±1σ) later. The lased spot 40Ar/39Ar data for phengite grains from the Mount Hamilton transition zone and rind suggest these grains began to crystallize 9 My before and continued to form 27 My (±1σ) after phengite grains crystallized in eclogite or garnet amphibolite. The vein in garnet amphibolite sample T-90-2A consists primarily of Ba-rich phengite grains with ages that overlap the much finer-grained phengite dispersed in the block itself (Table 2). LILE-rich fluids evidently mediated reaction between the exteriors of Tiburon garnet amphibolite or eclogite blocks and serpentinite to form actinolite- and phengite-rich rinds and transition zones between rinds and blocks for 2 to 12 My (±1σ). A Rb-Sr mineral isochron age of 152 ± 3 Ma for an epidote, chlorite, and white mica pod within a Tiburon eclogite block is interpreted as a minimum age for eclogite metamorphism (21). The result overlaps our age for the onset of LILE metasomatism at this locality. Because primary, low-salinity, aqueous fluid inclusions are preserved in clinopyroxene cores in both Franciscan and Samana eclogites (11, 22), these rocks were evidently metamorphosed in aqueous fluid-rich environments in events spanning tens of millions of years.

The discrepancy in ages for Tiburon versus Mount Hamilton high-grade blocks could reflect their relative proximities to LILE-rich fluid sources and the thermal histories of their host units. The Mount Hamilton eclogite block is from a structurally deeper unit than the Tiburon blocks (17). Phengite grains from its eclogite and garnet amphibolite layers cooled below their Ar closure T 2 to 26 My (±1σ) after, and average 0.5 wt % more BaO than, their counterparts from Tiburon blocks. The range of lased ages for single grains (Tables 2 and 3) implies heterogeneous, not thermally triggered, closure to diffusion. The Mount Hamilton block therefore probably underwent fluid-rock interaction for a longer time than the Tiburon blocks did, initially at higher temperatures, with a slower cooling history, and closer to the source of infiltrating phengite-forming fluids. However, deeper units in a subduction zone can be at the same (or lower) T than shallower ones, and variations of fluid compositions over time could yield lower BaO contents of phengite grains. An alternate explanation is that the BaO contents of these phengite grains were controlled by progressive fluid-rock interaction along varying fluid flow paths and that fluids of deep-seated origin had less access to shallower parts of the subduction complex with time.

Phengite 40Ar/39Ar ages link fluid-rock interaction to broader regional tectonics. The Eocene-Oligocene ages of grains from Samana eclogites resemble40Ar/39Ar ages of micas from eclogites in northern Venezuela (23); both may mark the transition from subduction to transcurrent uplift along the northern and southern margins of the Caribbean plate. The oldest spot ages for Franciscan Complex phengite resemble the 160 ± 3 Ma (16) and 160 to 170 Ma (24)40Ar/39Ar plateau ages for hornblende grains from Franciscan and Baja California amphibolite mélange blocks, which are interpreted to be “initiation of subduction” ages, whereas the youngest spot ages resemble values attributed to subduction zone metamorphism at lower P/T conditions (24). Our phengite-based chronology for Franciscan fluid flow also overlaps ages of voluminous magmatism in the Cordilleran Arc (25, 26). In our two paleosubduction settings, the grain-, sample-, and locality-scale age relationships of phengite within eclogite and related rocks document protracted time scales for LILE-rich fluid-rock interactions and link fluid flow to both large-scale tectonic and arc-magmatic events.

Supporting Online Material

Materials and Methods

Supporting Online Text

Figs. S1 and S2

Tables S1 to S9

  • * To whom correspondence should be addressed. E-mail: catlos{at}


Stay Connected to Science

Navigate This Article