Cu (R.A. Binns, unpubl. data), spanning the diagenesis and anchizone fields of Kübler (1964).
Key conclusions reached during Leg 193 regarding pervasively altered rocks underlying the thin cap of fresh lavas include the following (Binns, Barriga, Miller, et al., 2002):
In the following sections we review shipboard results and new research relevant to genetic issues, including unpublished petrologic and geochemical interpretations (R.A. Binns) necessary to fill certain gaps. A highly simplified longitudinal section from Snowcap to Roman Ruins illustrating salient aspects of the alteration system (Fig. F6) is useful for the discussions. The diagram depicts alteration below Sites 1188 and 1189 as parts of one large hydrothermal system below a cap of unaltered lavas with varying thickness (as suggested by deep-tow magnetic surveys), but the hypothetical continuity implied at depth is untested.
New clay mineral identifications in bulk rock samples measured by X-ray diffraction (XRD) and short-wave infrared spectrometry (SWIR) (Lackschewitz et al., this volume; Warden, this volume; Paulick and Bach, 2006) amplify shipboard data on the relative distribution of different clay mineral species below Sites 1188 and 1189. Specialized XRD investigations of clay mineral concentrates prepared from selected samples confirm the presence of occasional smectite accompanying chlorite between 34 and ~100 mbsf in Hole 1188A and clarify in particular the nature of varied mixed-layer phyllosilicates that may accompany illite and chlorite elsewhere (Lackschewitz et al., 2004). Although there are exceptions, mixed-layer clays appear more common in cristobalite-bearing assemblages and in rocks retaining relict plagioclase microlites.
Electron probe microanalyzer (EPMA) compositions of clay minerals are variously affected by fine grain size and intergrowths (Lackschewitz et al., 2004; Paulick and Bach, 2006; R.A. Binns, unpubl. data). Chlorites range from ripodolite to pycnochlorite (Mg/[Mg + Fe] = 45–85 mol%). EPMA data for illites demonstrate a wide compositional range from end-member illite (with 0.9 [K + Na] atoms per 11 oxygen anions, anhydrous, following Meunier and Velde, 2004) to lower-K variants containing substantial celadonite and pyrophyllite components. The ratio K/Na is variable, and some illites approach paragonitic compositions (Paulick and Bach, 2006, express uncertainty whether this represents solid solution or interlayering). Where measured, illites contain minor BaO (~0.2–0.4 wt%). A trend toward less sodic compositions with depth at Site 1188 occurs in EMPA and SWIR data (Warden, this volume; Paulick and Bach, 2006). This is the only systematic compositional trend with depth so far defined within the clay minerals. The illite crystallinity index measured following Kisch (1991) for clay fractions from 220 to 350 mbsf in Hole 1188F, and at 107 mbsf in Hole 1189A, range without a depth trend between 0.35° and 0.55°2Cu (R.A. Binns, unpubl. data), spanning the diagenesis and anchizone fields of Kübler (1964).
Chlorite and illite occasionally form thin seams in composite anhydrite veins and botryoidal growths or rosettes lining vesicles or irregular cavities, especially at Site 1188. Chlorite is absent from intervals of acid sulfate alteration and from pale selvages around fractures and anhydrite veins.
Comprehensive XRD and SWIR investigations (Lackschewitz et al., this volume; Warden, this volume) confirm two main pyrophyllite-bearing intervals below Snowcap, pervasive from 58 to 116 mbsf in Hole 1189A and more sporadic from 220 to 260 mbsf in Hole 1188F. The habit of pyrophyllite differs significantly between these intervals, as explained later. Pyrophyllite is unrecognized in Holes 1189A and 1189B under Roman Ruins.
Many samples from Site 1188 especially contain brownish patches and bands where clay minerals display distinctive waxy reflections in polished thin sections. The reason for this has not been established despite focused scanning electron microscope (SEM) studies, but the feature does not correlate with presence of pyrophyllite as suggested by Shipboard Scientific Party (2002b, 2002c). Distribution of the waxy brown patches commonly appears related to proximity of anhydrite veins or abundant disseminated anhydrite.
The relative distributions of different silica minerals in the subseafloor alteration system as established by shipboard XRD and optical microscopy have been confirmed by additional postleg identifications. Cristobalite and quartz are the main phases. Opal-A occurs within and immediately underlying the relatively unaltered volcanic cappings of Holes 1188A and 1189A, whereas rare XRD identifications of minor or trace tridymite in Hole 1188A are reported by Lackschewitz et al. (this volume).
Cristobalite tends to occur as tiny micrometer-scale particles and scattered rosettes intergrown with clays as alteration products of volcanic glass, and also as botryoidal linings to fractures and vesicles. It dominates assemblages in the domain above ~106 mbsf in Hole 1188A, whereas quartz dominates below ~106 mbsf and throughout Hole 1188F. The two species can occur in the same specimen throughout a transitional domain between 60 and 110 mbsf, which corresponds to the upper pyrophyllite interval of Hole 1188A. In Hole 1189A the transition from cristobalite to quartz is sharper and more shallow, at ~25 mbsf. Cristobalite also dominates approximately one-third of the lithologic units in the Lower Sequence (>118 mbsf) of Hole 1189B.
Quartz in altered wallrocks from the deeper domains of the alteration system is present as ragged anhedral grains 10–300 µm in diameter intergrown with clays (some resembling miniporphyroblasts), as euhedral crystals projecting into vesicles or filling amygdules, and as hairline veinlets. Continuous crystal growths or epitaxial overgrowths between these three habits imply close temporal relationship and at least local mobilization of SiO2. Silica phases are subordinate or absent within altered wallrock fragments in hydrothermal breccias from the Stockwork Zone of Hole 1189B and from some breccias or net-veined intervals in other holes. The breccia matrixes and vein networks associated with such wallrocks, however, are dominated by quartz.
Typically, the same silica phase present in altered rock also forms the partial linings or complete fillings to fractures and vesicles. In the transition intervals between cristobalite- and quartz-dominated domains some samples contain both phases. In these, cristobalite almost invariably occurs within the main rock body while the quartz occupies veinlets or fills former vesicles. From a variety of petrographic evidence such as pseudomorphous replacement textures it is clear that quartz generally formed later than cristobalite.
Igneous phenocrysts and microlites of calcic plagioclase are in places preserved within altered rocks from Sites 1188 and 1189 in association with either cristobalite or quartz. They are relatively common in the Lower Sequence of Hole 1189B. More generally they are altered to illitic clays.
Secondary sodic plagioclases are not commonly recognized outside the Lower Sequence of Hole 1189B (Paulick and Bach, 2006). Yeats et al. (2001) quote a compositional range from An5 to An15. Exceptional developments of granular secondary plagioclase are present in cavity-rich rocks from 175 to 185 mbsf toward the base of Hole 1188A, an interval where geochemistry denotes andesitic precursors and where some samples contain coexisting plagioclases close to pure albite and with An25 compositions, respectively, bridging the "peristerite gap" in low-temperature plagioclase solid solutions.
The presence of secondary potassium feldspar at Site 1189 has been established as quite widespread by postleg XRD, SEM, and EPMA studies (Yeats et al., 2001; Lackschewitz et al., this volume; Paulick and Bach, 2006). Potassium feldspar is conspicuously absent from altered rocks below Site 1188. In some samples from the Lower Sequence of Hole 1189B, potassium feldspar has replaced former plagioclase microlites. Elsewhere, tiny irregular to blocky grains, intimately intergrown with flakes of illite without reaction relationships, are clearly hydrothermal in origin. From limited microprobe analyses, the potassium feldspars contain significant hyalophane component (1–2 wt% BaO). The structural type of the potassium feldspar has not been established.
Anhydrite is the common sulfate phase disseminated within altered wallrocks and as vesicle fillings at Site 1188 but is by comparison subordinate at Site 1189. Barite is present in the same settings but is scarce and sporadic. Rare wallrock gypsum probably represents a retrograde hydration product of anhydrite. In vesicles, cavities, and veinlets, anhydrite typically crystallized late, after quartz or cristobalite.
Pyrite is the predominant disseminated sulfide in all alteration lithologies. It varies greatly in grain size and habit, ranging from anhedral to euhedral. Paragenetic observations regarding pyrite and other sulfide minerals are presented in the site chapters of Binns, Barriga, Miller, et al. (2002), but many aspects and especially the relationships between sulfide and silicate parageneses remain unclear. Chalcopyrite and sphalerite accompany wallrock and vesicle pyrite in some samples, especially at Site 1189 (Pinto et al., this volume). Pyrrhotite is exceptionally rare, and arsenopyrite has nowhere been observed in Leg 193 cores.
Hematite is a rare wallrock phase. Secondary magnetite is a scarce accessory in two main intervals of altered rocks below Snowcap, from 135 to 197 mbsf in Hole 1188A and from 318 to total depth (386 mbsf) in Hole 1188F. In the upper interval, dominated by andesite and mafic dacite precursors, fine-grained magnetite intergrown with clays appears part of the hydrothermal assemblage. In the lower interval, similar magnetite is restricted to the borders of unusual fractures occupied by clay minerals. Pinto et al. (this volume) identify ilmenite and hercynite accompanying magnetite in this latter interval and a single example of pyrophanite (MnTiO3). The lack of consistent compositional relationships for these coexisting phases precludes estimations of temperature or oxygen fugacity during alteration.
Rutile, apatite, and submicrometer-sized zircon are ubiquitous accessories intergrown with clays in altered wallrocks. Anatase occasionally substitutes for rutile (Lackschewitz et al., this volume). Microprobe data indicate that wallrock apatites are chlorine bearing, in contrast to apatites from anhydrite veins, which have high fluorine contents (Yeats et al., 2001). Other trace minerals with development restricted to unusual alteration styles include kaolinite, diaspore, brucite, talc, and alunite. Apart from rare tentative optical identifications, amphibole, epidote, and zeolites are conspicuously lacking. Although CO2 is a component of vent fluids at PACMANUS chimneys (20–40 mM in end-member hydrothermal fluid) (Ishibashi et al., 1996), calcite is known only as a trace mineral from 222 to 223 mbsf in Hole 1188F.
At Site 1188 there is a broadly consistent difference in microfabric between clay-dominated assemblages of the upper alteration domain characterized by cristobalite and the lower quartz-bearing domain. Many samples of the former were described on board ship as devitrified glass with undulose, plumose, or mesh-textured anisotropism. Analytical SEM investigations reveal exceptionally fine grained aggregates (grain size = 0.1 µm or less) of hydrothermal pyrophyllite or illite flakes, imperfectly to moderately well aligned (Fig. F7A). By contrast, in samples from the lower quartz-bearing domain the clay mineral flakes are coarser and loosely packed in random orientation (Figs. F7B, F7C, F7D). The coarser microfabrics dominate Hole 1189A and wallrocks in the Stockwork Zone of Hole 1189B, whereas in the Lower Sequence of Hole 1189B the two styles are interleaved according to the distribution of cristobalite and quartz. The spaces between the clay flakes in both styles amount to a considerable proportion of the rock volume and clearly explain the surprisingly high porosity of altered cores recognized in shipboard physical property measurements. Void spaces up to 70% by volume are estimated from SEM images.
New observations raise uncertainties with the concept advanced in Binns, Barriga, Miller, et al. (2002) of an initial, pervasive, chlorite-bearing alteration which became modified by subsequent silicification, acid sulfate alteration ("bleaching"), and development of pale selvages around fractures generally occupied by anhydrite veins. Shipboard interpretations of superimposed pervasive silicification were quickly discounted by onshore geochemical studies showing little change in SiO2 content relative to precursor lavas (see Fig. AF2 in the "Appendix"). Localized silicification, however, occurs on a millimeter to centimeter scale within wallrocks adjacent to certain but by no means all quartz veins and quartz-rich breccia matrixes in Holes 1189A and 1189B.
Figure F8A exemplifies macroscopic features from the upper pyrophyllite-bearing interval of acid sulfate bleaching in Hole 1188A that led Binns, Barriga, Miller, et al. (2002) to deduce late-stage superimposition of this alteration style. It illustrates an altered perlitic glass in which bleaching appears at first sight to be advancing from left to right as a replacement front facilitated by fractures. The dark greenish "kernels" consist of patchy mixtures of just-resolvable illite, mixed-layer phyllosilicate, cristobalite, disseminated euhedral pyrite, and minor anhydrite. The white portion of the sample has scattered cristobalite grains within optically irresolvable, pyrophyllite-dominated "glass" with undulose to mesh-textured anisotropism and is actually broken up by the anhydrite-cristobalite-pyrite–filled fractures rather than being developed as a reaction selvage around them. No replacement or pseudomorphous structures suggesting dissolution of kernel-style pyrite are present in the pale material. Since reversion of the apparently more crystallized clay assemblage of the dark kernels to an "anisotropic glass" appears most unlikely, the structure is more easily interpreted if the pyrophyllitic bleaching (acid sulfate alteration) were in fact the first phase of alteration or, alternatively, if the two alteration styles were essentially simultaneous.
Samples from the lower pyrophyllite-bearing interval in Hole 1188F vary from uniformly pale in color to mottled, and many are not obviously associated with anhydrite veining. Analytical SEM studies reveal the mottling to reflect variations in the modal proportions of illite and relatively coarse (1–3 µm) pyrophyllite. Where both occur, adjacent flakes of illite and pyrophyllite coexist in apparent equilibrium. Pyrophyllite thus appears here to be more definitely part of a primary alteration assemblage.
Figure F8B illustrates a simple variant of pale selvage beside an anhydrite vein, the third kind of superimposed alteration considered by Binns, Barriga, Miller, et al. (2002) as a product of wallrock reaction between anhydrite vein-forming fluid and chloritic wallrock. Such selvages typically have a sharp border against a faintly layered "transition zone" that passes gradually into the inner, chloritic "kernel." These structures are largely restricted to Site 1188 and are more common in the quartz-bearing domain, especially in Hole 1188F. More complex examples have multiple-layered selvages suggestive of repeated diffusion fronts. Not uncommonly, plagioclase microlites or phenocrysts are preserved in the kernel portions but are altered to clays in the selvage and transition zone, often abruptly, again apparently suggesting later development of the selvages. Greater abundance of disseminated anhydrite in the pale selvage zone supported a genetic relationship to the anhydrite vein.
Reexamination of many samples from Holes 1188A and 1188F reveals that there is not an exact correspondence between anhydrite veins and the pale selvages. Figure F9 schematically presents observations that collectively indicate the anhydrite veins were emplaced after formation of the pale selvages. Whereas the selvages clearly developed from fluids transported via preexisting fractures, the anhydrite veins fill openings caused by reactivation of the fractures and a jumbling or rotation of wallrock blocks between them.
Described as pyrophyllitic by Shipboard Scientific Party (2002b), most pale selvages in Holes 1188A and 1188F are in fact composed largely of illite intergrown with quartz (or cristobalite) and disseminated anhydrite (Fig. F7C). Pyrite is characteristically lacking or very scarce. In the greenish kernels illite is intergrown with chlorite or Mg-bearing mixed-layer clays (Fig. F7B), pyrite is common, but anhydrite is absent or subordinate. In the transition zones, chlorite generally becomes less abundant relative to kernels without evidence of replacement by illite, anhydrite may increase, and pyrite is usually less common but may be enriched in particular laminae. In the only known example of a selvage containing pyrophyllite, from the mottled lower pyrophyllitic interval of Hole 1188F, this mineral is also present (with smectite-illite mixed-layer phyllosilicate) in the kernel.
Considering the exceptional porosity of clay mineral assemblages, the sharp outer borders of most selvages seem inconsistent with an origin by diffusive replacement of formerly chloritic assemblages, although the layering in transition zones certainly recalls double-diffusion processes. Where pyrite first appears abundantly in transition zones it shows no signs of corrosion or replacement. Pseudomorphous structures indicating former presence of kernel-type pyrite replaced by quartz or illite are altogether lacking in the selvages, and in some samples contrasted quartz microfabrics between selvages and kernels also appear inconsistent with replacive overprinting.
Although features such as the distribution of disseminated anhydrite remain to be explained (it would need to replace the clay minerals), the petrographic evidence again seems more consistent with simultaneous alteration of precursor volcanic wallrock in response to progressive composition changes in inwardly permeating fluids or, alternatively, that the selvages represent an earlier alteration phase with kernels altering later under changed fluid compositions.
Regardless of these genetic issues, petrologic studies permit a simplified classification of alteration types (Table T1), presented to facilitate the following discussion of alteration chemistry. It covers formerly coherent volcanic rocks. Rare or unusual alteration styles, as in some volcaniclastic horizons or flow-banded and spherulitic rocks, are not included.
Assuming immobility of Ti and Zr to assess precursor compositions, Lackschewitz et al. (2004) evaluate chemical changes in five altered samples from Holes 1188A and 1188F. Some but not all of the bulk elemental changes observed were explicable in terms of clay concentrate compositions. From a more extensive study of both Sites 1188 and 1189, Yeats (2003, 2004) notes that bulk Si is generally conserved, Mn is universally depleted, K is especially enriched at Site 1189 in rocks containing hydrothermal K-feldspar, and S is typically enriched by 2–3 orders of magnitude. A number of chalcophile elements (Zn, Pb, As, and Sb) display enrichment in the upper section of Hole 1188A but become progressively depleted at depth and in Hole 1188F; Zn returns to near-precursor levels at the base of Hole1188F. Copper and Mo were considered to be enriched throughout Holes 1188A and 1188F, but Yeats (2004) cites "chaotic" behavior of chalcophile elements in Holes 1189A and 1189B.
These conclusions do not take account of significant differences in composition between adjacent samples with varied alteration style (Table T2), which could obscure genetically significant downhole or cross-site trends. In order to assess this, major and trace element data for the Commonwealth Scientific and Industrial Research Organisation (CSIRO) collection of Leg 193 samples have been examined in terms of the petrologic classification of Table T1. Procedures for calculating precursor rock compositions are outlined in the "Appendix." The principal uncertainty is with Cu, which exhibits significant variability and an abrupt change in fractionation behavior within Pual Ridge lavas (Sun et al., 2004). Salient outcomes of the study follow (see the "Appendix" for further details):
For some elements the compositions of high-temperature fluids venting at Satanic Mills and Roman Ruins (Gamo et al., 1996; Douville, 1999), and also the computed hydrothermal components of borehole fluids (Binns et al., this volume), are consistent with potential leaching from the drilled volcanic sequence during hydrothermal alteration. For example, Li and Mn are consistently highly depleted (relative to precursors) in altered rocks and correspondingly enriched (relative to seawater) in the vent fluids. However, this does not apply to many other elements, notably potassium. High K/Ca ratios in vent fluids characterizing PACMANUS fluids in contrast to basalt-hosted sites elsewhere were ascribed by Gamo et al. (1996) to reaction with felsic rocks. However, K enrichment is typical of most alteration styles at Site 1188 and particularly Site 1189. Only the pyrophyllitic bleaching (Category XB) of Hole 1188A has the potential to release significant K, and this style of alteration is lacking below Roman Ruins.
Overall gains and losses during alteration, based on averages for all CSIRO samples irrespective of alteration style and possible nugget effects are listed in Table T3 for major subdivisions at Sites 1188 and 1189. Although representivity is uncertain, the data emphasize, for example, the pronounced loss of SiO2 from wallrocks in the Stockwork Zone of Hole 1189B and the marked differences in Mg, K, Ba, and U behavior between the two sites. High copper enrichments in a few samples at Site 1188 outweigh Cu depletions in others, leading to apparent overall gains within both cristobalite and quartz domains that may be statistically questionable. Overall Cu gains are more definite in Hole 1189A and the Lower Sequence of Hole 1189B. Zn appears lost overall from Site 1188 and from Category Z wallrock fragments in the Stockwork Zone of Hole 1189B, but it is gained overall in Hole 1189A and more so in the Lower Sequence of Hole 1189B. Upward transport of Pb is a possible explanation for the contrast between the enriched cristobalite domain and the depleted quartz domain at Site 1188, but at Site 1189 the wallrocks are again enriched overall in absolute terms. The behavior of gold has not yet been studied.
The elevated porosities of altered rocks measured on board ship have been confirmed on shore by Christiansen and Iturrino (this volume). They note that samples with >5% of plagioclase microlites from Site 1188 have relatively constant porosity (~15%–25%), but porosities of totally altered samples fall with increasing depth from ~30% below the fresh lava cap to ~15% at 336 mbsf. At Site 1189, Category Z wallrocks in the Stockwork Zone, lacking a silica phase, are exceptionally porous (45%–65%).
Sufficient chemically analyzed samples have measured bulk densities, or densities that can be reasonably inferred from equivalent neighbors, to allow assessment of volume changes during hydrothermal alteration of formerly hyaline lavas at PACMANUS (Fig. F10). Assumed immobility of Ti and Zr is used to determine precursor compositions (see the "Appendix") and thereby precursor densities from measurements on fresh Pual Ridge glasses and microlitic rocks.
Site 1188 shows a pronounced drop in volume expansion from ~60% directly below the fresh lava cap in Hole 1188A to ~20% at 150 mbsf, which continues as a more modest decline to ~15% at ~360 mbsf in Hole 1188F. The profile is interrupted by an interval of exceptionally high expansion (up to 92%) between ~170 and ~200 mbsf, the zone of cavity-rich altered andesites (Category A and a related Y-K). Apart from these there are no consistent differences in the depth relationships between the various alteration categories. Volume expansions in Holes 1189A and 1189B range from low values to 76%, with less systematic downhole trends.
Total volume change as plotted in Figure F10 represents an interplay between induced porosity (dilation arising from excess fluid pressure), volume changes governed by alteration reactions (cf. Tenthorey and Cox, 2003), and mass gains or losses. These factors may have additive or opposed influence. For example, the Category Z altered perlite at 30.12 mbsf in Hole 1189B is a case where there has been marked overall expansion during alteration (by 58%), yielding a rock with 67% porosity in which the solid constituents now have a bulk density of 3.26 g/cm3 and have shrunk to 54% of their original volume. From this and similar calculations for other samples it is evident that the induced porosity factor dominates.
The mineralogy, geometry, abundance, and apparent timing relationships of veins from 1 mm to several centimeters wide at Sites 1188 and 1189 are extensively described by Shipboard Scientific Party (2002b, 2002c). The dominant vein assemblage at Site 1188 is anhydrite ± (pyrite) ± (cristobalite or quartz). An early generation of pyrite often forms marginal laminae at the wallrock contact. The mineralogy, structure, and paragenesis of this vein style are very similar to those of vesicle and amygdule fillings, and indeed there are examples of continuous development where veins cut into large vesicles. Observations presented above on the relative timing of illite selvages remove a main basis on which a sequence of veins at Site 1188 was deduced by Scientific Party (2002b).
At Site 1189, similar anhydrite veins are present but distinctly less frequent in Hole 1189A and in the Lower Sequence of Hole 1189B. Those in Hole 1189A typically contain more quartz than those at Site 1188 and may cut across an earlier generation of quartz veins, some with minor chalcopyrite and sphalerite (Shipboard Scientific Party, 2002c).
Hydrothermal breccias are also common, ranging from close-packed, jigsaw-fit structures to matrix-supported breccias with dispersed fragments of altered wallrock. The matrixes, variously dominated by anhydrite or quartz, generally with some pyrite, are essentially equivalent to vein fillings. In the Stockwork Zone of Hole 1189B the distinction between veins and breccias within the small samples recovered is essentially the proportion and relative dispersal of altered wallrock fragments. Prismatic quartz and euhedral pyrite favor margins against altered wallrock, whereas late-stage anhydrite may occupy the centers of veins or drusy pockets within breccia matrixes. Pyrite varies from a minor to a major constituent. In the matrixes of occasional hydrothermal breccias in Hole 1189A, anhydrite again typically has a late-stage paragenesis relative to quartz.
At the base of the Stockwork Zone (118.1 mbsf) a coarse monomineralic anhydrite vein cuts directly across a set of quartz-dominated veins containing pyrite, chalcopyrite, sphalerite, minor anhydrite, and the only recorded occurrence of gold (Pinto et al., 2003, 2005, this volume). Nearby, bladed anhydrite in the matrix of a rare second-generation breccia cuts earlier breccia fragments with quartz-pyrite matrixes.
A close relationship between Category Z wallrocks (depleted in SiO2) and quartz-veined stockworks suggests that the source of SiO2 in certain quartz veins and breccia matrixes may be relatively local. Wallrock fragments picked free of veins in a finely net-veined interval at 106.5 mbsf in Hole 1188A lost 260 g of SiO2 per kg of precursor, but a bulk sample of the interval shows overall conservation of the precursor levels. In many samples from the quartz domain of Hole 1188F where bulk SiO2 content was unchanged during alteration there is evidence of internal redistribution into abundant vesicle fillings, discontinuous veinlets, and large grains resembling porphyroblasts.
Barriga et al. (2001, 2004) ascribe formation of some breccias, particularly variants with stylolite-like residual enrichments in fine-grained rutile/anatase at the borders of atypically rounded fragments, to local corrosion or dissolution during hydrothermal alteration followed by collapse of remaining clasts. However, the determination that alteration predominantly involves volume expansion induced by overpressured fluids favors dilational hydrofracturing as the predominant process.
Few pieces of core recovered during Leg 193, all from Site 1189, contained sufficiently abundant sulfides to classify as massive (>75% sulfide) or semimassive (25%–75%) in ODP terminology. Nor were thick sulfide bodies detected by wireline or resistivity-at-the-bit logging at either Site 1188 or Site 1189 (Bartetzko et al., 2003). Small core fragments originally logged as massive sulfide from the Stockwork Zone of Hole 1189B represent portions of pyrite-quartz veins or breccia matrixes (Binns, this volume).
Two core samples of semimassive sulfide represent the only significant intersections recovered. An 8-cm piece of sulfide rock from 31.0 mbsf in Hole 1189B contains irregular massive aggregates or nodules of coarse pyrite with chalcopyrite overgrowths, set in a dark gray matrix of bladed to granular anhydrite (partly altered to gypsum) with disseminated pyrite and chalcopyrite. Amoeboid pods of pale gray, bladed anhydrite largely devoid of sulfides represent a final stage of deposition in what appears to have been a progressively dilated deposit characterized by open-space mineral growth. Barite is lacking, and quartz and sphalerite are rare. Fragments of altered dacite with disseminated sulfides are scattered throughout the sample, which thus appears a variant of breccia from the Stockwork Zone. Similar chalcopyrite-bearing "nodular breccias" were cored by CONDRILL shallow drilling (<5 mbsf) in the vicinity of Hole 1189B (Petersen et al., 2003, 2005; Binns, 2005). Although containing barite as well as later anhydrite, these samples also carry dispersed fragments of totally altered volcanic wallrock and they too have structures indicative of successive dilation and open space mineral growth. Together with the semimassive sulfide at 31.0 mbsf in Hole 1189B, these are interpreted as upper levels of the Stockwork Zone, underlying rather than merging above into the mound of collapsed chimneys at Roman Ruins.
The other large semimassive sulfide sample, at 107.7 mbsf in Hole 1189A, occurs within a layered volcaniclastic paleoseafloor horizon. In a band ~5 cm thick, abundant sulfides (predominantly pyrite with lesser chalcopyrite and sphalerite) together with quartz and minor anhydrite form a matrix to well-dispersed fragments of altered tube pumice and nonvesicular volcanic rock. In places the matrix invades altered volcanic clasts along fractures. Above and below the semimassive sulfide band are altered hyaloclastites with similar clasts, here closely packed and self-supported but containing only minor matrix sulfide. Both contacts of the ~5-cm layer are serrated as if the matrix of the sulfide layer was invading the bordering hyaloclastites.
Although contrary interpretations are offered by Shipboard Scientific Party (2002c) and Paulick et al. (2004), the semimassive sulfide in Hole 1189A is considered a mineralization variant formed during the same event that created stockwork veins and semimassive sulfide nearby in Hole 1189B. Preservation of angular outlines of the delicate tube pumice particles indicates that mineralization involved dilation of the deposit rather than widespread replacement. Equivalent seabed hyaloclastites with tube pumice particles, collected by dredge and submersible on Pual Ridge, are moderately well sorted and close-packed, as expected from submarine spallation of lava rinds (Waters et al., 1996; Waters and Binns, 1998).
Pinto et al. (this volume) provide electron probe microanalyses of pyrite, chalcopyrite, and sphalerite from the semimassive sulfides. Binns (this volume) presents bulk compositions and isotopic data for five samples of massive and semimassive sulfide and for a hammer-drill sand (containing subequal proportions of altered wallrock and sulfide fragments) derived from the 0- to 30-mbsf cased portion of Hole 1189B. Maximum Cu and Zn concentrations are 5.0 and 1.6 wt%, respectively. Gold contents are moderate (0.9–2.0 ppm). The subseafloor semimassive sulfides are not as rich in base and precious metals as chimneys at Roman Ruins (average Cu = 7.0 wt%, Zn = 24 wt%, Au = 16 ppm, and Ag = 230 ppm) (Binns, 2004) but are richer in Co, Te, and Bi and are mildly enriched in Se. Isotopic ratios (Pb and S) of the subseafloor samples and seafloor chimneys are similar.
Problems involved with estimating temperatures for hydrothermal and low-temperature metamorphic clay mineral assemblages arising from disequilibrium and sensitivity to fluid compositions (Árkai, 2002; Meunier and Velde, 2004) are compounded by the extrapolations involved in applying experimental data to the pressures of the subsurface PACMANUS hydrothermal system drilled during Leg 193 (0.16–0.20 kbar). The general presence of chlorite and illites with crystallinity indexes spanning the diagenesis and anchizone fields, plus comparisons with alteration in geothermal fields, suggests temperatures of 220° to 300°C, below those of the zeolite or greenschist facies of metamorphism (Browne, 1978; Srodon and Eberl, 1984; Árkai, 2002). Presence of pyrophyllite rather than kaolinite with silica minerals in alkali-poor assemblages at these relatively low pressures, by extrapolation from experiments of Chatterjee et al. (1984), favors the upper part of that temperature range.
Assuming equilibration with seawater-like hydrothermal fluid, calculated temperatures based on oxygen isotope ratios in near-monomineralic concentrates of illite and illite-dominated mixed-layer illite-smectite from 233 to 355 mbsf in Hole 1188F range from 250° to 300°C, increasing downhole (Lackschewitz et al., 2004). An illite at 40 mbsf in the Stockwork Zone of Hole 1189B has a calculated temperature of 305°C. Chlorites from illite-poor samples at 175 mbsf in Hole 1188A (225°C), 116 mbsf in Hole 1189A (250°C), and 118 and 197 mbsf in the Lower Sequence of Hole 1189B (235° and 220°C) yield temperatures lower than those for illites, whereas pyrophyllite concentrates between 50 and 117 mbsf in Hole 1188A yield elevated temperatures (260°–310°C).
No oxygen isotope temperature estimates have been made for pyrophyllite-free alteration in the cristobalite domain at Site 1188. Presence of smectite in some samples from this domain suggests alteration took place at least partly at temperatures lower than the 225°C calculated for chlorite at 175 mbsf.
Oxygen isotope ratios measured on quartz samples from veins, breccia matrixes, and vesicle linings in Holes 1189A and 1189B yield mostly higher temperature estimates, from 280° to 400°C (Lackschewitz et al., 2004; R.A. Binns and A.S. Andrew, unpubl. data). Because an empirical fractionation relationship extrapolated from low-temperature cherts (Knauth and Epstein, 1976) was used, it is not clear whether these apparently higher formation temperatures relative to those of alteration clay minerals are meaningful, but this is supported by the following fluid inclusion data.
Studies of fluid inclusions in vein anhydrite at Sites 1188 and 1189 unequivocally establish that phase separation occurred below the seafloor (Vanko et al., 2004; Vanko and Bach, 2005). Trapping temperatures range from 150° to 320°C at 48 mbsf. The range decreases progressively and temperatures increase at depth, to 270°–375°C below 200 mbsf. Knowing that anhydrite only forms in seawater above 150°C, the lower trapping temperatures correspond closely to the present-day thermal gradient interpolated from a borehole temperature measurement of 313°C at 360 mbsf (Shipboard Scientific Party, 2002b). The uppermost trapping temperatures mostly coincide with the two-phase boiling curve for seawater at the appropriate hydrostatic pressures (Bischoff and Pitzer, 1989). Most calculated salinities are close to seawater values, but some approach pure water and others are extremely saline (to 31 wt% NaCl equivalent). Trapping temperatures at Site 1189 are uniformly high (242°–368°C) and range as high as those of the seawater boiling curve. Salinities range from near pure water to ~7 wt% NaCl equivalent.
Preliminary studies of fluid inclusions in hydrothermal quartz from veins, breccia matrixes, and amygdules at Sites 1188 and 1189 yield results very similar to those in anhydrites, including evidence of phase separation (Vanko et al., 2006). At Snowcap, quartz trapping temperatures between 230 and 340 mbsf range from 300° to 400°C. Between 60 and 155 mbsf at Roman Ruins they range from 267° to 380°C. Although anhydrite tends consistently to be paragenetically later than quartz, the two minerals evidently formed at similar temperatures.
No temperature estimates based on isotope ratios or fluid inclusions are available for the two intersections of semimassive sulfide at Site 1189, nor have the illite selvages of Hole 1188F been compared with associated kernels in this respect.
Glassy dacite lava belonging to a sequence that forms the substrate for active sulfide chimneys at Tsukushi has demonstrably flowed over seabed outcrops of altered rhyodacites at Snowcap and contains a xenolith of altered rock. This unequivocal evidence of overlap between hydrothermal activity and volcanic eruptions during the more recent history of Pual Ridge raises the question of whether cycles of volcanic eruption and hydrothermal activity were repeated many times during the construction of the Pual Ridge edifice.
Several observations at Site 1189 suggestive of repeated volcanic and hydrothermal events, initially recorded by Shipboard Scientific Party (2002c), are used by Paulick et al. (2004) in formulating their model of three phases of volcanism and hydrothermal activity. At ~186 mbsf in the Lower Sequence of Hole 1189B, chalcopyrite and sphalerite are present in an altered perlite clast within a polymict volcaniclastic horizon but are apparently absent from the matrix or other clasts, suggesting to Paulick et al. (2004) erosion of an outcrop mineralized during a hydrothermal event at the end of their Phase 1. Within their Phase 2, differently colored clasts with varying proportions of chlorite, illite, and quartz in another polymict volcaniclastic horizon at 117 mbsf in Hole 1189A are considered to be derived by erosion from previously altered outcrops with contrasted alteration styles. The overlying mineralized volcaniclastic horizon with tube pumice clasts (107.7 mbsf) is considered a possible fossil seafloor or immediately subseafloor hydrothermal deposit by Paulick et al. (2004). At Site 1188, noting that the upper pyrophyllite-bearing alteration interval in Hole 1189A could arise from trapping of highly acid fluids below the impermeable cap of unaltered lava at Snowcap, Paulick and Bach (2006) suggest the lower pyrophyllitic unit in Hole 1188F might represent similar ponding beneath a former but no longer unaltered impervious cap present at the end of Phase 2 (225 mbsf, placed at a boundary between sparsely porphyritic and aphyric dacite by Paulick et al., 2004). These observations have other possible explanations and need further testing.
The arguments supporting the opposite hypothesis—that within the drilled section of Pual Ridge there has been only one overall hydrothermal and subsurface mineralization event—include
Systematic downhole profiles for the ranges in fluid inclusion trapping temperatures for vein anhydrites and quartzes at Site 1188 imply one overall veining event throughout Holes 1188A and 1188F, imposed certainly after construction of most of the Pual Ridge edifice but not necessarily implying a single alteration event if that were earlier.
We consider that there is no convincing evidence of multiple alteration episodes in the cores at either Site 1188 or Site 1189, even though this is conceptually possible as shown by the "unconformable" Tsukushi-group flow at Snowcap. Clearly, however, within the alternative single event there have been complex variations in fluid compositions and perhaps physical conditions from place to place and over time. The pervasive alteration systems below both sites are large relative to the areal extent of seafloor hydrothermal activity. If they indeed link at relatively shallow depth as drawn in Figure F6, the subseafloor PACMANUS alteration system is an exceptional feature requiring copious fluid and heat flux. Abundance of unstable glass in the volcanic sequence may have reduced the energy requirements.
The Lower Sequence of Hole 1189B does not exhibit such regular alteration profiles. Its intercalated cristobalite-bearing and quartz-bearing assemblages might at face value suggest repeated alteration events, but from overview of cores with the benefit of reasonably good recovery Shipboard Scientific Party (2002c) inferred one event with varied alteration styles controlled by precursor lithology. Cristobalite-bearing rocks are dominated by coherent facies that were less permeable to hydrothermal fluid flow, whereas the quartz-bearing units were typically of brecciated and flow-banded facies.
Cristobalite is not a stable silica polymorph at hydrothermal alteration temperatures (Heaney, 1994). However, it commonly forms outside its stability conditions during burial diagenesis, in low-temperature wallrock alteration at geothermal sites, and in acid sulfate alteration halos around some epithermal ore deposits. In active geothermal fields, Browne (1978) notes that cristobalite is generally restricted to temperatures below 100°–115°C, whereas quartz occurs at higher temperatures. He suggests that temperature is the dominant control, although differences in silica activity of geothermal fluids modify the relative distribution. Jones and Segnit (1972) proposed that if cristobalite nuclei are present (derived, for example, by high-temperature devitrification of volcanic glass), then epitaxial precipitation of cristobalite may be energetically more favorable in hydrothermal systems than deposition of quartz. Dissolution-precipitation experiments at 150°–300°C (Renders et al., 1995) confirm that cristobalite precipitates from hydrothermal solutions if Si(OH)4 concentrations exceed cristobalite saturation but are below quartz saturation and provided cristobalite nuclei are present. Except for the fact that the temperatures suggested by Browne appear unrealistically low for the entire cristobalite domain at Site 1188, the scenario is an attractive one for PACMANUS. Relict high-temperature cristobalite formed by devitrification is more likely to be present in felsic volcanic rocks than in basalts, and this could explain the contrast between PACMANUS and the Trans Atlantic Geotraverse (TAG, Mid-Atlantic Ridge) hydrothermal site, where chalcedony and quartz rather than cristobalite dominate subseafloor altered assemblages (Hopkinson et al., 1999).
A variety of petrographic evidence in the broader transitional zone between the two domains at Site 1188 indicates that quartz generally formed later than cristobalite, locally replacing it. Hence, a time factor is also involved, in which respect the boundary between cristobalite and quartz domains (shown schematically on Fig. F6) is not the equivalent of a metamorphic "isograd" arising from a conductively imposed thermal gradient preceding or contemporary with alteration once fluid flow commenced. Instead, upwelling of progressively hotter hydrothermal fluids evidently converted earlier cristobalite-dominant assemblages to quartz-bearing assemblages at the sharp transition of Hole 1189A and across a broader domain in Hole 1188A. Advective heating by the fluids at Site 1188 and most of Site 1189 thereby established an overall but irregular upward thermal gradient, complicated by small-scale cells or eddies. Interleaving of cristobalite-bearing products of coherent-facies lavas and quartz-bearing brecciated facies in the Lower Sequence of Hole 1189B may arise from kinetic factors related to fluid/rock ratios together with or instead of temperature differences; lateral movement of quartz-forming fluid along permeable breccia horizons may outweigh effects of vertical upwelling.
Accepting that the altered rocks drilled at Sites 1188 and 1189 represent a single overall event, many variations in fluid compositions and in the nature of fluid/rock interactions were involved, the relative timing of which remains subject to uncertainty and requires focused inquiry.
New observations suggest that the pale illite-rich selvages, formed at Site 1188 from exceptionally potassic fluids low in Mg and Fe, represent an early phase of alteration governed by preexisting fractures, perhaps primary igneous cooling features. Alteration then followed of remnant glassy "kernels" to more chloritic assemblages (with illite and pyrite) through reaction with relatively reduced Mg-rich fluids. Preservation of igneous plagioclase reflects fluid composition rather than partial alteration; fluids associated with plagioclase-destructive alteration were relatively more enriched in potassium and reduced sulfur. Permeation of fluids from late-stage anhydrite veins into adjacent country rock appears to have had only limited effects, such as growth of disseminated anhydrite in porous illite selvages and development of a "waxy clay" phase on the surfaces of preexisting illites.
In the upper part of Hole 1188A, zones of pervasive pyrophyllite-bearing acid sulfate alteration ("bleaching") represent reaction with particularly low pH fluid deficient in both Mg and K. The timing of this style of alteration relative to development of kernel-style chloritic assemblages is unclear. Microscopic characteristics suggest an early status for acid sulfate alteration, in conflict with macroscopic appearances. Alternatively, pervasive formation of pyrophyllitic bleached assemblages and associated chloritic kernel assemblages was contemporaneous and dictated by local variations in fluid chemistry. At deeper levels of Site 1188, pyrophyllite is texturally in equilibrium with illite, favoring the contemporaneous alteration alternative.
Limited quantitative modeling (Yeats et al., 2001; Paulick and Bach, 2006) indicates low-pH (<3), low-[K+]/[H+] fluids associated with pyrophyllitic acid sulfate alteration at Site 1188 vs. higher-pH (>5), high-[Mg++]/[H+] fluids associated with chloritic kernel assemblages. Only extremely rarely was pH high enough to stabilize carbonate minerals. Chloritic assemblages containing hydrothermal potassium feldspar at Site 1189 reflect higher fluid [K+]/[H+] than in equivalents lacking this phase at Site 1188, rather than different temperatures. Rare alunite-bearing assemblages such as at 346 mbsf in Hole 1188F require exceptionally acid fluid (pH < 2).
Alteration temperatures estimated for the drilled portion of Pual Ridge are insufficient for phase separation, although in vein anhydrites and quartz there is clear evidence for this process quite close to the seafloor, possibly after most alteration had been effected. More deeply seated phase separation during the alteration event, however, provides a potential genetic link between low-pH and higher-pH alteration fluids plus a mechanism for widespread hydrofracturing and brecciation. Faster uprise of an aggressive vapor phase and condensed derivatives potentially explain the petrographic indications that development of pale selvages to fractures came first, preceding alteration of unaffected glass "kernels" to chloritic assemblages by reaction with higher-pH, Mg-bearing seawater or brine derivatives. Further support for such processes comes from laser ablation microanalyses of sulfur isotopes in pyrite, discussed later.
Given the paucity of massive and semimassive sulfide mineralization with elevated base and precious metal contents within the pervasively altered hydrothermal system drilled during Leg 193, scant direct evidence is available for assessing temporal and chemical relationships between hydrothermal alteration of the Pual Ridge volcanic sequence and the formation of sulfide chimneys at the seafloor. From geochemical reaction modeling, Yeats et al. (2001) and Paulick and Bach (2006) observe that hydrothermal end-members for vent fluids collected from Satanic Mills chimneys (Douville et al., 1999) are in equilibrium with pyrophyllite + quartz or kaolinite + cristobalite. End-member compositions and pH of vent fluids collected at 180°C from Roman Ruins (e.g., molar K/Ca = 7.8) are similar to those of fluids collected at higher temperatures (250° and 268°C) from Satanic Mills (molar K/Ca = 5.0) (Gamo et al., 1996; T. Gamo, pers. comm., 1996). It is accordingly perplexing that no equivalent pyrophyllite- or kaolinite-bearing alteration occurs in either Holes 1189A or 1189B, both of which were collared beside chimneys. Instead, the common presence of K-feldspar and chlorite (together with illite) in alteration assemblages at Site 1189 indicates alteration fluids with substantially different chemical properties from those forming the chimneys. Together with the disparity between K enrichment of altered rocks as well as of vent fluids at Site 1189, this suggests that the exact conduits for present-day chimney-forming fluids at Roman Ruins were either not intersected or not recovered during Leg 193, for at least some diagnostic acid alteration would be expected in their vicinity.
The relatively few samples recovered in the Stockwork Zone of Hole 1189B, combined with those of CONDRILL shallow drilling, show a vertically upward change from quartz veins to quartz-anhydrite veins and breccia matrixes to nodular semimassive sulfides into a fallen chimney mound then the active chimney field. Pyrite dominates the sulfide assemblages except in the chimneys, whereas chalcopyrite then sphalerite become increasingly abundant in semimassive sulfides then dominate the chimneys. Drawing comparisons with fossil volcanogenic massive sulfide systems, this sequence certainly suggests a genetic relationship between sulfides in the Stockwork Zone and the seafloor chimneys. However, altered wallrocks (Category Z) from the Stockwork Zone are conspicuously potassic and denote equilibration with near-neutral fluids rather than those currently venting at seafloor chimneys. The same difficulties apply to rare quartz veins, carrying minor chalcopyrite, sphalerite, and even gold, that are cut by the more common style of quartz-anhydrite vein at Site 1189.
Overall, much of the extensive subseafloor alteration evident at PACMANUS may have been completed prior to uprise of high-temperature acid fluids in constrained (but as yet undiscovered) conduits feeding the chimneys.
One of the more striking outcomes of research on Leg 193 cores has been evidence of dilational processes during formation of the subseafloor hydrothermal system at PACMANUS. Preservation of open or partly filled vesicles and cavities in altered volcanic rocks, many of which lie buried beneath considerable overburden, is a remarkable feature of the drilled sequence. Geochemical modeling and density measurements indicate substantial volume expansion of the volcanic sequence during hydrothermal alteration, especially in the upper cristobalite domain. The dominant factor in this expansion is an imposed dilation arising from elevated pore fluid pressures (greater than lithostatic), creating the high porosity between loosely packed clay mineral particles. This creates a most fertile environment for diffusion and mineral deposition, something that would not be recognized in metamorphosed ore environments where the physical evidence will not survive.
Dilational fracturing and infill of created spaces with hydrothermal minerals (quartz, anhydrite, and pyrite) are characteristic of hydrothermal breccias and most veins at Sites 1188 and 1189. The hydrofracturing process again requires excess fluid pressure. Dilation is also ascribed importance for formation of the two semimassive sulfide intersections below Roman Ruins and in shallow CONDRILL equivalents. Subsurface sulfide deposition in open spaces created by such dilation is a process deserving consideration for so-called "subhalative" or "subseafloor exhalative" massive sulfide deposits in ancient sequences where the textural evidence may be destroyed by metamorphism or incorrectly interpreted as replacement wallrock silicate minerals.
One process that could cause repeated dilation is subsurface phase separation or boiling of hydrothermal fluids. If transported rapidly enough in freshly created fractures to avoid cooling by conduction, batches of fluid close to their boiling point at depth would flash on rising adiabatically to a lower-pressure environment resulting in explosive fragmentation followed by mineral deposition. It is unlikely that boiling occurred during pervasive alteration of the drilled volcanic sequence because estimated alteration temperatures are well below the boiling curve of seawater at 1600–2000 mbsl. However, the evidence for boiling fluids in vein anhydrites and quartz is unequivocal (Vanko et al., 2004). Excess fluid pressures during alteration could be a consequence of boiling at deeper levels, transmitted hydraulically through pores and fractures. Because of latent heat effects, flashing of uprising fluid batches will be short lived. Displacements on tectonic fractures could cause more dramatic but perhaps less frequent flashing and dilational effects.
A possible large-scale effect of expansion of the hydrothermal alteration system could be fracturing and breaching of the carapace or cap of relatively impermeable, unaltered lava, thereby enabling escape and venting of fluids at the seafloor (e.g., at Roman Ruins). In this respect, the overall PACMANUS system might be visualized as a geological "pressure cooker."
On a smaller scale, excessive dilation of delicately cohesive clay aggregates might lead to fluidization, an alternative to the process of hydrothermal corrosion and collapse proposed by Barriga et al. (2001, 2004) as a causal mechanism for some breccia structures and for generation of incoherent domains in the Stockwork Zone of Hole 1189B that might in future become the locus of a large subhalative sulfide ore body.
