IGNEOUS PROCESSES

Thermobarometry

Our synthesis paper (Kelemen et al., submitted [N1]) concentrates on three observations:

  1. Compositions of plutonic rocks and primitive mid-ocean-ridge basalt (MORB) samples from the 14°–16°N region along the Mid-Atlantic Ridge are indicative of relatively high pressure crystal fractionation in a 15- to 20-km-thick thermal boundary layer.
  2. Most gabbroic rocks sampled during Leg 209 are too evolved to be complementary to crystal fractionation of MORB and may not have a volcanic equivalent. Instead, they may represent crystallization products of melts that solidified entirely in the subsurface, with no volcanic expression.
  3. There is a marked lack of crystal shape fabric or other evidence of penetrative subsolidus deformation of residual mantle peridotites recovered during Leg 209, along with a high proportion of ductile shear zones and faults at each site, indicative of localized deformation during seafloor spreading and tectonic denudation of mantle rocks within this 15- to 20-km-thick thermal boundary layer.

In this section, we concentrate on the evidence for igneous crystallization at relatively high pressure.

As reported in the Leg 209 Initial Results volume (Kelemen, Kikawa, Miller, et al., 2004), unusual poikilitic peridotites from Site 1275 had "equilibrated" textures prior to low-temperature alteration (Fig. F3). Additional illustrations are available in Shipboard Scientific Party (2004a: fig. F51; 2004g: figs. F6, F18, F40). The poikilitic peridotites contain olivine, orthopyroxene, clinopyroxene, plagioclase, and Cr-rich spinel. Their whole-rock molar Mg/(Mg+Fe), or Mg#, and Ni are high, as are Mg# and Ni contents of olivine and molar Cr/(Cr+Al), or Cr#, of spinel, extending to residual peridotite values. However, plagioclase ranges from 54 to 75 mol% anorthite (Fig. F4), much lower than expected in residual peridotites, particularly because Leg 209 residual peridotites are among the most depleted recovered from the mid-ocean ridges (e.g., Shipboard Scientific Party, 2004f: fig. F51 and accompanying text). The wide range of plagioclase anorthite content in the poikilitic peridotites from Site 1275, at nearly constant olivine Mg# and Ni content, is best understood as the result of "reactive fractionation" involving crystallization during reaction between refractory peridotite and cooling, fractionating melt migrating along olivine grain boundaries. In such processes, Mg# and Ni content are "buffered" by exchange reactions with Mg, Fe, and Ni-rich olivine, whereas the abundance of incompatible elements such as Na—which preferentially partition into melt versus crystals—increases with decreasing melt mass (e.g., Kelemen, 1986).

In Figure F4, note that gabbroic rocks from very slow spreading ridges with thin crust (e.g., Cayman Rise) have the lowest anorthite content in plagioclase at a given olivine Mg#, whereas gabbroic rocks from the fast-spreading East Pacific Rise, sampled at Hess Deep, have the highest anorthite content for a given olivine Mg#. This variation has commonly been interpreted to indicate crystallization from primitive magmas with a variety of Na/Ca ratios, in turn reflecting a variety of different degrees of melting at different mantle potential temperatures, with high-Na magmas arising from low degrees of melting and low-Na magmas from high degrees of melting (e.g., Meyer et al., 1989). However, an alternative interpretation is that primitive Cayman magmas with high Na content arose as a result of extensive "reactive fractionation" in a thick thermal boundary layer, whereas Hess Deep magmas with much lower Na contents underwent little or no "reactive fractionation" in the shallow mantle, perhaps because the thermal boundary layer beneath the fast-spreading East Pacific Rise is very thin compared to that beneath the slow-spreading Cayman Rise.

Variation in the intensity of "reactive fractionation" has the potential to account for much of the variation in the sodium content of primitive mid-ocean-ridge basalts, which has hitherto been attributed to variations in mantle potential temperature. This hypothesis forms the basis for ongoing research (Collier and Kelemen, 2005, 2006). However, this work is not yet published, and we forego a more extensive summary here.

The pressure and temperature of equilibration between melt and a mineral assemblage including olivine, two pyroxenes, and plagioclase can be estimated from the melt composition using the method of Kinzler and Grove (1992). Rare earth element (REE) contents in Site 1275 poikilitic impregnated peridotites (Kelemen et al., submitted [N1]) are consistent with crystallization from "normal" MORB. Kelemen, Kikawa, Miller et al. (2004) and Kelemen et al. (submitted [N1]) used the composition of 87 primitive MORB glasses at 14°–16°N with Mg# 0.60–0.73 (from the Petrological Database of the Ocean Floor, PetDB, online at petdb.ldeo.columbia.edu/petdb). These primitive MORB glasses could be olivine + two pyroxene + plagioclase saturated at 0.54 GPa (±0.14 GPa, 2) and 1220°C (±16°C, 2) (Shipboard Scientific Party, 2004a: fig. F53 and accompanying text). Other proposed melt composition barometers yield similar results (Kelemen et al., submitted [N1]).

A variety of metamorphic reactions could occur during cooling and decompression of olivine–two pyroxene–plagioclase–spinel mineral assemblages. To the extent that minerals approach equilibrium during this process, thermodynamic data for minerals in these reactions can be used to estimate the pressure and temperature conditions for "closure," after which metamorphic reactions effectively ceased. The boundary between plagioclase and spinel lherzolite facies (e.g., Green and Hibberson, 1970) is defined by the reaction

MgAl2O4 + Mg2Si2O6 + CaMgSi2O6 = 2 Mg2Si2O4 + CaAl2Si2O(1)
MgAl2O4 in spinel ss + enstatite in orthopyroxene + diopside in clinopyroxene = 2 forsterite in olivine + anorthite in plagioclase

that relates the activities of the most abundant mineral end-members in the olivine, orthopyroxene, clinopyroxene, and plagioclase solid solutions in our samples. Reaction 1, together with temperature estimates from independent geothermometers, has been used to estimate pressure in terrestrial and lunar olivine + 2 pyroxene + plagioclase + spinel assemblages (e.g., Frost, 1976; McCallum and Schwartz, 2001; Newman et al., 1999; Tartarotti et al., 2002).

To estimate closure conditions for Reaction 1 in impregnated peridotites from Site 1275, Kelemen et al. (submitted [N1]) used mineral solution models and internally consistent thermodynamic data for most mineral end-members from Holland and Powell (1998) and the Sack and Ghiorso (1991) solution model for Cr-Al-Fe-Mg spinels. An independent estimate of temperature was derived from the Lindsley and Anderson (1983) two-pyroxene solvus thermometer, as implemented in the computer program QUILF (Andersen et al., 1993). Two-pyroxene solvus temperatures based on clinopyroxene compositions, together with analytical uncertainty, range from ~1075° to 1110°C (with one sample representing an outlier at ~1215°C) ± 50°C. Other pyroxene solvus thermometers yielded consistent results, essentially independent of pressure. Combining these temperature estimates with Reaction 1, incorporating analytical uncertainty, yields pressures of 0.65 to 0.72 ± 0.1 GPa (1).

In summary, Kelemen et al. (submitted [N1]) concluded that metamorphic net transfer equilibria involving MgAl2O4, Mg2Si2O6, CaMgSi2O6, 2 Mg2Si2O4, and CaAl2Si2O8 in impregnated peridotites from Site 1275 last equilibrated at pressures of ~0.68 ± 0.2 GPa and temperatures of 1100° ± 75°C. This estimate is consistent, within uncertainty, with isobaric cooling from the estimated igneous crystallization conditions for primitive MORB from this region, described above and in Kelemen, Kikawa, Miller, et al. (2004). Impregnated peridotites and olivine gabbronorites at Sites 1268, 1270, and 1271 contained olivine, two pyroxenes, and plagioclase, ± spinel, prior to alteration. They have similar whole-rock compositions to the impregnated peridotites at Site 1275 and so probably record similar conditions of crystallization.

Results of thermobarometry support the inference that igneous rocks crystallized and began to conductively cool at depths 15–20 km below the seafloor in the 14°–16°N region along the Mid-Atlantic Ridge. This result is consistent with previous estimates of the thermal boundary layer thickness beneath slow-spreading ridges based on thermal models (Braun et al., 2000; Reid and Jackson, 1981; Shen and Forsyth, 1995; Sleep, 1975). It is also consistent with geological inferences based on dredging and diving in the 14°–16°N region (e.g., Cannat, 1996; Cannat and Casey, 1995; Cannat et al., 1997) and with interpretation of the liquid lines of descent formed by spatially and genetically related mid-ocean-ridge basalts, which indicate cotectic crystallization at pressures of 0.4–0.6 GPa along parts of the Mid-Atlantic Ridge (Elthon, 1993; Grove et al., 1992; Meurer et al., 2001; Michael and Chase, 1987), including the 14°–16°N region (Xia et al., submitted [N2]). More generally, this result may also be consistent with suggestions that crystallization of cooling melt migrating by porous flow can modify the composition of many upper mantle peridotites beneath slow-spreading ridges (e.g., Dick, 1989; Niu, 1997).

Gabbroic Rock Compositions

As noted above, Kelemen et al. (submitted [N1]) also emphasize the evolved nature of most of the gabbroic rocks recovered during Leg 209. As shown in Figure F5 (modified from Shipboard Scientific Party 2004a: fig. F58), primitive MORB has an Mg# ~ 0.7 and ~50 ppm Zr, whereas average MORB has an Mg# ~ 0.6 and ~100 ppm Zr. If Zr were a perfectly incompatible element (concentration in crystal/liquid = 0), then doubling of Zr concentration would require 50% crystallization. In fact, Zr is not perfectly incompatible (concentration in crystal/liquid > 0), which suggests that average MORB is the product of >50% crystal fractionation. There should be a complementary reservoir of refractory high-Mg#, low-Zr cumulates formed by this crystal fractionation. Such a refractory cumulate reservoir is observed in the southern massifs of the Oman ophiolite (e.g., Pallister and Hopson, 1981). However, refractory cumulates are rare among gabbroic suites sampled from mid-ocean ridges. The average compositions of gabbroic rocks from ODP Hole 735B on the Southwest Indian Ridge (Natland and Dick, 2002) and ODP Leg 153 sites on the Mid-Atlantic Ridge (Agar and Lloyd, 1997) are similar to those of primitive MORB and thus cannot represent the refractory cumulate reservoir. Similarly, gabbroic plutons sampled during Leg 209 at Sites 1270, 1272, and 1275 are far too rich in iron to explain the variation trend of MORB.

Only primitive gabbronorites at Site 1268 and impregnated peridotites at Sites 1268, 1270, 1271, and 1275 have appropriately high Mg# and low Zr to represent the refractory cumulate reservoir required by MORB variation. As noted above, these rocks, with their high Mg# and abundant orthopyroxene, probably formed at depths of 15 km or more. We infer that the more evolved gabbroic rocks sampled during Leg 209 formed at lower temperatures and shallower depths and thus that crystal fractionation is polybaric in the 14°–16°N region. Many of the more evolved gabbroic rocks are similar to MORB glass compositions and may not be cumulates at all. Additionally, some gabbroic rocks from Leg 209 are so evolved that they have no volcanic equivalents among MORB. We infer that a substantial proportion of the melts entering the thermal boundary layer in the 14°–16°N region crystallize entirely at depth, without any eruptive products on the seafloor.

Impregnation of Residual Peridotites

Seyler et al. (2007) discuss the origin of interstitial clinopyroxene in residual peridotites from Leg 209, commonly with symplectitic textures involving chromian spinel (e.g., see figures and associated text in Shipboard Scientific Party, 2004a: fig. F43; 2004b: figs. F10, F11; 2004c: fig. F24; 2004f: figs. F17–F22). Clinopyroxenes are also intergrown with magmatic sulfides whose composition and texture are presented for the first time in Seyler et al. (2007). These unusual clinopyroxenes are associated with anomalously high CaO contents at a given Al2O3 concentration (e.g., Shipboard Scientific Party, 2004f: fig. F51 and associated text).

Similarly, orthopyroxenes in residual peridotites from Leg 209 exhibit interstitial textures, apparent association with embayed olivine crystals, and coarse sympletitic intergrowths with spinel (see figures and associated text in Shipboard Scientific Party 2004b: figs. F6, F7, F9, F12, F44–F46, F77, F91; 2004c: figs. F19, F28, F38, F91; 2004d: fig. F42; 2004f: figs. F10–F16).

Using new major and trace element analyses of silicates and new analyses of sulfides, Seyler et al. (2007) conclude that the association of clinopyroxene, spinel, and sulfide minerals was produced by reaction between residual orthopyroxene and a melt saturated with respect to clinopyroxene and sulfur. The introduction of this melt into previously depleted residual peridotites produced these intergrowths and changed the bulk composition of the rocks. It is increasingly evident that many, if not most, residual peridotites recovered during dredging of the mid-ocean ridges have undergone some extent of "refertilization" by a process similar to that proposed by Seyler et al. (2007). For example, in a recent landmark paper, Niu (2004) showed that most dredged peridotites contain enrichments of highly incompatible elements that are not reflected in residual mineral compositions and yet involve fluid-immobile elements such as Nb, Ta, and Th, so probably result from addition of a late igneous phase along residual crystal grain boundaries, without equilibration with grain interiors.

Takazawa et al. (this volume) contributed a study of ultramafic and mafic rocks from Hole 1271B. Shipboard observations document the lithologic assemblage of predominantly dunite, with lesser brown amphibole gabbro, olivine gabbro, and troctolite, and a minor amount of harzburgite and olivine gabbronorite. The interpretation of the probable genesis of this assemblage was initial crystallization of plagioclase and clinopyroxene from basaltic liquid migrating along grain boundaries. As the section cooled the flow regime changed from porous media to fracture-focused flow, resulting in crystallization of mafic rocks in an ultramafic host. The results reported in Takazawa et al. (this volume) are consistent with this interpretation. They note a variance in Fo content of olivine between dunite (~89–91) and mafic rocks (~86–89) and a concomitant variability in chrome spinel compositions. This is attributed to more refractory olivine and spinel in the dunite owing to melt-rock reaction as the dunite represents a fossilized melt channel.

Metamorphism and Hydrothermal Alteration

Bach et al. (2004) reported on petrographic observations of alteration in peridotites from Leg 209, supplemented by thermodynamic modeling of fluid/rock reactions. Petrographic data in the paper summarize and add to extensive shipboard observations, made largely by the same group of authors (see the "Metamorphic Petrology" sections in each site chapter in Kelemen, Kikawa, Miller, et al., 2004). Serpentine and talc ± amphibole are observed and predicted in relatively SiO2 rich residual harzburgites, whereas serpentine and brucite are observed and predicted in SiO2-poor dunites. Alteration reactions controlled in large part by these bulk compositions probably controlled fluid pH, fO2, and concentrations of dissolved species.

A notable exception to the rock-dominated alteration assemblages observed at most sites is the extensive replacement of early formed serpentine by talc at Site 1268. Some residual peridotite whole-rock compositions determined onboard the JOIDES Resolution were identical to stoichiometric talc, despite the fact that textures clearly indicated that the protoliths were olivine-rich, porphyroclastic, and protogranular harzburgites. Although it has long been known that some altered peridotites have low Mg/Si compared to unaltered protoliths, it is debated whether this is due to Mg removal (e.g., Snow and Dick, 1995) or SiO2 addition (e.g., Boschi et al., 2006).

In the Site 1268 summary (Shipboard Scientific Party, 2004a: figs. F12, F13, and associated text), we argued that wholesale replacement of peridotite by talc at essentially constant Mg/Fe almost certainly involved addition of SiO2 (adding ~40 wt% SiO2 relative to the original rock mass). If, instead, the change was produced mainly by removal of Mg and Fe (removing at least 16 wt% of the original rock mass), it would be a dramatic coincidence for this process to preserve the original Mg/Fe ratio of the protolith. Bach et al. (2004) support our conclusion in the site summary, noting that the solubility of Mg in hydrothermal fluid is very low in the presence of hydrous Mg-Fe silicates at pH > 6, which is in the pH range that they calculate likely for fluids during serpentinization and subsequent talc alteration at Site 1268. However, they hedge their bet and invoke the need for further study. Doubtless, studies of sulfide mineralogy and, particularly, stable isotope thermometry on samples from this site are likely to shed further light on the conditions before and during talc alteration.

Pursuing the notion that SiO2 was indeed added to harzburgites to produce the talc-rich rocks, Bach et al. (2004) suggest that the source of SiO2 could have been the abundant gabbronorite intrusions into peridotite that were sampled at Site 1268. Some Site 1268 gabbronorites have indeed undergone metasomatism, most notably loss of CaO (Shipboard Scientific Party, 2004a: fig. F14C and associated text). However, in the Site 1268 summary we showed that SiO2 contents in altered gabbronorites from Site 1268 are essentially identical to SiO2 in fresh gabbronorite (Shipboard Scientific Party, 2004a: fig. F14), suggesting that the extent of SiO2 leaching from gabbroic intrusions was minor compared to the extent of SiO2 addition to peridotites. Furthermore, in some gabbronorite samples from Site 1268, orthopyroxene is pseudomorphically replaced by talc (Shipboard Scientific Party, 2004b: fig. F30 and associated text), which requires addition of SiO2 to the pyroxene pseudomorphs. Thus, it seems unlikely that the gabbronorites that were actually sampled at Site 1268 were the source of the SiO2 added to the talc-rich peridotites. Since gabbroic intrusions comprise <50% of the volume of core from Hole 1268A, we can infer that SiO2 loss from the gabbronorites did not balance SiO2 gain in the peridotites. In this view, the source of SiO2 metasomatically added to the Site 1268 talc-altered peridotites has yet to be identified.

Paulick et al. (2006) interpret whole-rock major and trace element characteristics of peridotites recovered during Leg 209 in terms of residues of partial melting, igneous "impregnation," and metasomatism during hydrothermal alteration. Following Niu (2004), they ascribe variable light REE concentrations in samples from Site 1270 and 1271, not correlated with heavy REE concentrations, to igneous impregnation because the light REEs are correlated with concentrations of fluid-immobile high field strength elements and Th. This is in agreement with shipboard interpretation of thin section relationships and major element chemistry in terms of addition of igneous, "cumulate," interstitial plagioclase, and clinopyroxene, with or without hornblende, zircon, and other minor phases to depleted harzburgite and to plagioclase- and pyroxene-free dunite during porous flow of cooling, crystallizing melt in the shallow mantle beneath the Mid-Atlantic Ridge.

In contrast, Paulick et al. (2006) interpret light REE variation that is weakly correlated with both heavy REE and high field strength element concentrations in residual peridotite samples from Sites 1268, 1272, and 1274 to metasomatism during hydrothermal alteration. Within the peridotites from Sites 1268, 1272, and 1274, Paulick et al. (2006) find that "rock-dominated" serpentinization controlled the composition of most Site 1274 samples, with addition of H2O and oxidation but little additional metasomatism during shallow melt migration or alteration. This may be somewhat at odds with the evidence for impregnation of igneous clinopyroxene in Site 1274 peridotites described by Seyler et al. (2007), whose results are summarized above.

Alteration strongly affected the composition of residual peridotites from Site 1268. Paulick et al. (2006) extensively describe the effects of "fluid-dominated" serpentinization at Site 1268, including "gains in sulfur and development of [a] U-shaped REE pattern with strong positive Eu anomalies." High-temperature fluids (350°–400°C) with these trace element characteristics are observed in hydrothermal vents (Paulick et al., 2006).

A third type of metasomatism during hydrothermal alteration, as delineated by Paulick et al. (2006), is associated with the replacement of olivine, pyroxene, and serpentine by talc at Site 1268, yielding smooth light REE–enriched trace element patterns with negative Eu anomalies.

Bach et al. (2006) reiterate observations on serpentinization parageneses and textures from the Leg 209 Initial Reports volume (Kelemen, Kikawa, Miller, et al., 2004), emphasizing that zones with comparatively high water-rock ratios (mesh rims and veins) contain serpentine plus magnetite, whereas zones with lower water-rock ratios (mesh cores and rims of relict olivine) contain serpentine plus brucite. High water-rock ratios result in both hydration and oxidation of the peridotite protolith, whereas olivine-rich protoliths with low water-rock ratios show only hydration. Both of these processes are potentially isochemical with regard to all species other than H2O and oxygen. Bach et al. (2006) link these observations with shipboard measurements of magnetic properties and density; oxidized samples with abundant magnetite are denser and have more pronounced magnetic susceptibility than unoxidized samples containing brucite and no magnetite.

Moll et al. (this volume) report microprobe analyses of a variety of primary and alteration phases in peridotite samples from Leg 209. Primary minerals at Site 1274 are very "depleted" with high Fo contents in olivine and high Cr/Al in spinel, as anticipated from whole-rock compositions determined during Leg 209. As was clear during the cruise (e.g., Shipboard Scientific Party, 2004f: fig. F51 and accompanying text) and reiterated by Seyler et al. (2007)—described above—and Harvey et al. (2006)—summarized below—these are among the most depleted residual peridotites known along the mid-ocean ridges. In samples where pseudomorphs of olivine and orthopyroxene may be clearly distinguished, Moll et al. (this volume) analyzed alteration phases replacing these minerals. Although there are some differences between serpentines replacing olivine compared to those replacing pyroxene, the similarities are striking and suggest that major element cations were extensively redistributed during alteration. It is clear that the authors plan to publish a more complete analysis of the chemical changes associated with alteration at a later date, and one can anticipate that this will be very interesting.

Sulfide mineral compositions from Hole 1268A are reported by Miller (this volume). Sulfide mineral species change downsection from millerite and chalcopyrite in shallow cores to pyrrhotite and pentlandite with depth. The general downhole trend suggests sulfide mineral precipitation in conditions with decreasing sulfur and oxygen fugacity. This trend is interrupted coincident with cores that recovered a brecciated gabbroic intrusion. Sulfide minerals that indicate precipitation at relatively higher sulfur and oxygen fugacity as well as sphalerite occur in the central core of the intrusion breccia. Strongly contrasting pyrite compositions suggest at least two episodes of pyrite precipitation, but there is no clear morphological distinction between phases.

NEXT