Since the inception of the Deep Sea Drilling Project (DSDP) and its successor, the Ocean Drilling Program (ODP), one of the principal objectives of the science community has been to penetrate an entire section of oceanic crust to reach the crust/mantle boundary. This objective has proven to be a difficult engineering task from the initial efforts of ocean drilling to the present. Attempts to initiate and drill holes in young, highly fractured basaltic rock with little or no sediment cover have proven unsuccessful (e.g., Legs 106, 109, 142). However, efforts to initiate deep basement holes with considerable sedimentary cover have proven successful in yielding more stable holes. For example, Hole 504B reached a depth in excess of 2100 m before the hole was lost to further penetration. Approximately two-thirds of the crust below, however, remained unsampled, as did much of the drilled crust because of low recovery.
Following this and other attempts at total crustal penetration, new strategies were adopted based on successful drilling in tectonic windows into peridotite during Leg 109 along the Mid-Atlantic Ridge near the Kane Transform (MARK Area; Detrick, Honnorez, Bryan, Juteau, et al., 1988) and into gabbroic rocks during Leg 118 near the Atlantis II Transform on the Southwest Indian Ridge (SWIR; Robinson, Von Herzen, et al., 1989). This new approach involved total crustal penetration by a strategy called offset drilling, in which tectonically exposed lower crustal and upper mantle sections were targeted. By drilling multiple long holes laterally offset in the same region, different levels of the crust and upper mantle could be sampled. Using these laterally offset holes presumably would allow a composite section of the oceanic crust to be constructed, once correlated. Subsequently, this approached was used during Legs 147 (Hess Deep; Gillis, Mével, Allan, et al., 1993) and 153 (MARK Area; Cannat, Karson, Miller, et al., 1995) and again during Leg 176 (Atlantis II Transform; Dick, Natland, Miller, et al., 1999).
With the exception of Hole 735B drilled during Leg 118 and 176 (1508 m of total penetration), which was initially spudded using a guide base on an highly unusual wave-cut platform along the Atlantis II Fracture Zone in the Southwest Indian Ocean, other ODP igneous legs conducted in young terrane, where there is little to no sedimentary cover, have been successful only in penetrating and coring to about 200 meters below the seafloor (mbsf) or less. They have also been characterized by far less recovery than experienced during Legs 118 and 176. Experience gained from the two legs that directly followed the inception of offset drilling strategy, Legs 147 (Hess Deep; Gillis, Mével, Allan, et al., 1993) and 153 (MARK region; Cannat, Karson, Miller, et al., 1995), indicates that the current hard-rock guide base design is not optimal for establishing boreholes in fractured igneous rock environments with moderate slopes and little or no sediment cover. These are typical environmental conditions for mid-ocean ridge and fracture-zone drilling targets. Thinly sedimented slopes that are covered with debris or rubble are also commonly encountered in young terrane at mid-ocean ridges. These areas have proven especially problematic from an engineering point of view because of the difficulties in keeping a stable open hole for deepening.
Even with these formidable engineering and technical problems, the scientific objective of total crustal penetration or developing a composite section by drilling offset holes in the ocean crust remains one of the highest priority thematic objectives of ODP. Therefore, new hardware and techniques need to be developed to establish boreholes in these environments to meet the scientific objectives of hard-rock drilling. The tool with the most promise of dramatically increasing ODP's ability to establish a borehole in hard-rock environments is the hammer drill-in casing system (HDS). Thorough testing of this tool during Leg 179 before its general deployment in an actual hard-rock leg should increase the likelihood of success in the future. Therefore, the engineering portion of Leg 179 was dedicated solely to testing a hammer drill-in casing system in a fractured hard-rock environment. Tests were conducted along a transverse ridge adjacent to the Atlantis II Transform fault, which offsets the Southwest Indian Ridge. The hammer-drill test sites are located near the highly successful Hole 735B (Fig. F1). This region was chosen for the tests because of its well known geological framework and the fact that it possesses an uncommon combination of hard-rock drilling targets, including shallow to deep-water exposures in both flat and highly sloped environments.
Drilling and coring operations in fractured hard rock must overcome many challenges not confronted in piston coring operations. These can be summarized as initiating the borehole, stabilizing the borehole, and establishing reentry capability. Until a drilling/coring bit can gain purchase, because it is not stabilized by sediment, it tends to chatter across the surface of a hard-rock outcrop. Difficulty initiating a hole is exacerbated if the drilling target is on a slope. Rubble from the seafloor, drill cuttings, and material dislodged from the borehole wall must continually be removed; however, the size and density of this material complicates this task. Because of bit wear in hard rock, deep penetration (beyond a few tens of meters) absolutely requires the ability to perform multiple entries into a borehole. The ideal system for drilling in hard-rock environments would disregard local topographic variation, seafloor slope, and thickness of sediment cover or talus accumulation. Such a system should initiate a hole, then simultaneously deepen the hole and stabilize the upper part of the hole with casing. This requires the bit to cut a hole with a greater diameter than the casing and then to be withdrawn through the casing string. The casing in turn would facilitate hole-cleaning operations by elevating the annular velocity of the drilling fluid and would ease reentry operations by eliminating the possibility of offsets in the borehole wall (ledges or bridges). Finally, this ideal system would leave behind a structure to simplify the required multiple reentries.
The hammer drill-in casing system is composed of a hydraulically actuated percussion hammer drill, a casing string or multiple casing strings, a free-fall deployable reentry funnel, and a casing hammer. Once the casing string has been drilled into place and the reentry funnel installed, the drilling assembly is unlatched from the casing string and removed. The borehole is left with casing and a reentry funnel in place. If required, the casing string may be cemented in place, and multiple casing strings may be installed in the same borehole.
This type of HDS is currently being
used in Iceland to install large diameter 18
-in casing more than 100
m deep in fractured basalt. Unfortunately, the Icelandic system is pneumatically driven
and, thus, is not suited for use in deep water. However, a hydraulically actuated hammer
drill, suitable for use by ODP, is currently under development in Australia. ODP is
assisting in the development of this hammer drill and has incorporated it into the HDS.
A viable HDS would (1) eliminate the need for any form of independent seafloor structure such as the hard-rock base, (2) allow spudding boreholes on much steeper slopes than can be achieved using an independent seafloor structure, (3) reduce sensitivity to thin sediment cover, debris, or rubble lying on the spudding surface, and (4) reduce dependency on precise site surveys.
ODP initiated its HDS project in
1994 with a worldwide industry survey of the available hammer-drill technology,
techniques, and equipment. In July 1996, ODP was invited to visit an Iceland Drilling
Company drill site where 11
-in casing was being drilled into fractured
basalt using a pneumatic hammer drill. It was determined that similar techniques could be
employed by ODP. However, because of the water depths typically associated with ODP legs,
the pneumatic hammer drill would have to be replaced with a hydraulic, or water, powered
hammer drill. Further industry surveys resulted in locating SDS Digger Tools, Canning
Vale, Western Australia, which had a 6-in prototype water-powered hammer drill that was
ready for commercialization. Discussions with SDS Digger Tools resulted in an agreement
between the company and ODP to work together to scale up the existing 6-in water hammer to
a size suitable for drilling in 16-in casing.
To test the general concept, in August 1996, a field test of the existing SDS Digger Tools 6-in water hammer was carried out. The field test was successful in drilling 7-in casing into black granite in a quarry. Since SDS Digger Tools was not in the business of making underreaming hammer-drill bits, in September 1996 the decision was made to use underreaming hammer-drill bits manufactured by Holte Manufacturing, Eugene, Oregon, U.S.A. Holte Manufacturing has been in the business of drilling in casing into hard fractured rock for many years, in many locations around the world, using pneumatic hammer drills.
In October 1996, SDS Digger Tools
presented the option to ODP of using an existing prototype 12
-in water hammer. The 12-in
water hammer could be used to drill in 13
-in casing and would
cost less to complete development than developing an entirely new hammer capable of
drilling in 16-in casing. Therefore, the decision was made to change the prototype HDS
from 16-in casing to 13-in casing and to employ the SDS Digger Tools
prototype 12-in water hammer. In January 1997, ODP engineers traveled to Perth,
Australia, to witness bench testing of the prototype SDS 12-in water hammer. The
bench test was successful, and the project was continued based on the 12-in
water hammer drill.
The 12-in water hammer was
field tested in black granite in April 1997. Although the field tests, from a drilling
standpoint, were successful, it was determined that the hammer closing forces were too
high for safe operation from the drillship. A redesign of the 12-in
water hammer was undertaken by SDS to lower the closing forces. A second round of field
tests was carried out with a modified 12-in water hammer in
September 1997 at Rogaland Research Center, Stavanger, Norway. The results of the second
field test indicated that the closing forces were now in an acceptable range for use by
ODP.
During development of the 12-in
water hammer drill by SDS Digger Tools, ODP/TAMU (Texas A&M University) developed the
supporting hardware required for the HDS system. This hardware included a hydraulically
actuated casing running tool, a modified 13-in casing hanger, a
bearing assembly between the modified casing hanger and casing string, and a free-fall
reentry cone. The bearing assembly between the modified casing hanger and casing string
was added to allow the drilling assembly to rotate independently of the casing string. The
free-fall reentry cone was designed to be assembled around the drill pipe and dropped to
the seafloor, coming to rest on the modified 13-in casing hanger.
Besides making reentry easier, the free-fall reentry cone locks out the bearing between
the casing hanger and casing string. Locking out the bearing is required for installation
of other casing strings using conventional ODP casing running tools, which must be rotated
to latch and release. The HDS running tool, hanger bearing, and free fall-reentry cone
were assembled and fit tested at ODP/TAMU in March 1998. All of the HDS equipment was
shipped to Cape Town, South Africa, in April 1998 for testing at sea during Leg 179.
There were three primary objectives for the hammer drill-in casing evaluation. The first objective was to determine the operational characteristics of the hammer drill. The hammer drill was thoroughly tested on land before it was deployed at sea; however, it was difficult to simulate the shipboard deployment environment. Thus, during Leg 179 the hammer was deployed by itself for evaluation prior to using the entire hammer drill-in casing system. The second phase of testing would involve determining the viability of the hammer drill-in casing system. Once the shipboard operational characteristics of the hammer drill were established, the complete hammer drill-in casing system would be deployed for evaluation. Three boreholes in increasingly difficult environments were envisioned to completely test the equipment. Finally, our third test phase would involve determining the maximum allowable slope for hammer-drill operations. The proposed drilling plan addressed the minimum requirements to evaluate the potential of a hammer drill-in casing system. No coring was specifically planned; however, we did envision the possibility of attempting recovery of at least two cores through the established boreholes.
Initially, we planned to deploy the hammer drill on top of the wave-cut platform. Our projected site, based on the Leg 118 seafloor survey, was in an area of very thin (<1-2 cm) sediment cover ~75 m west of Hole 735B in ~730 m of water. Once shipboard operational parameters were determined, we intended to assemble and deploy the entire hammer-drill system and attempt to set and recover, if possible, 40-60 m of casing string at the same location. Pending successful completion of the casing installation at our first site, we planned to establish cased holes in deeper water and eventually on a sloped surface.
The sea trials for the hammer drill-in casing system took place near Site 735 along the Atlantis II Transform adjacent to the SWIR (Fig. F1). There were some difficulties in maintaining the original 15-day schedule for the hammer-drilling tests. First, the JOIDES Resolution was delayed before leaving port because of a major repair to the drill string guide horn damaged in a previous leg. The ship departed Cape Town on 21 April, more than 6 days late. Second, a delayed shipment of hammer bits and drill-in casing supplies had not reached the ship before the JOIDES Resolution left Cape Town. These materials had to be sent to the Resolution while on station near Hole 735B, but the resupply ship did not arrive until 10 May, ~8 days after supplies on hand for the hammer-drilling tests were exhausted. Thus, before the hammer supplies arrived, only 3 days of hammer tests were completed at Site 1104 before hammer supplies were exhausted (see "Operations"). The scheduled 15 days of tests had to be suspended in order to await transfer of the parts by a resupply ship from Reunion Island. A contingency plan had been developed in the event of catastrophic failure of the HDS. Implementation of this contingency plan included rotary coring time at Site 1105, which lies ~1.2 km east-northeast of Site 735B on the Atlantis Bank. This unexpected result of the leg resulted in extensive coring in the gabbroic massif exposed near Site 735B. Because of lost operational days and a further failure in transfer of all materials from the resupply ship because of agitated seas, serious compromises were made to the HDS testing and planned experiments, as well as to drilling originally scheduled for the Ninetyeast Ridge Observatory (NERO) site on the second part of the leg. In all, ~ 17 days of operational time out of an original 26 days devoted to the primary objectives of the leg (hammer tests and NERO) were lost as a consequence of these delays. The hammer tests were resumed on 11 May, but only for a 2-day period at Site 1106, where three alternate bit designs were tested (see "Operations").
Hole 735B, the hammer test sites (Sites 1104 and 1106), and Site 1105 lie on or adjacent to the Atlantis Bank. This platform is located to the east of the north-south-trending Atlantis II Transform (Figs. F1, F2), which offsets the SWIR in a left-lateral sense by a distance of 199 km (Engel and Fisher, 1975). The transform lies between 31º50'S and 33º40'S along ~57ºE. The age offset is ~20 Ma, and the slip rate along the transform is ~0.8 cm/yr (Fisher and Sclater, 1983; Dick et al., 1991c). Because of the slow spreading rates of the SWIR segments to the north and south, the rift valleys adjacent to the transform and the transform valley itself are characterized by strong relief.
Dick et al. (1991b) presented the first detailed multibeam map of the Atlantis II Transform and adjacent ridge-transform intersections (RTIs). The multibeam map of the transform and a 3-D shaded relief image (Figs. F2, F3) are reproduced here based on the original site-survey digital data (kindly provided by H. Dick). These figures display the major morphotectonic features of the Atlantis II Transform. As originally described by Dick et al. (1991c), these features include two RTIs, with well-developed nodal basins, inner corner highs, and more subdued outer corners of the rift valley; the transform valley; transverse ridges east and west of the transform valley; and two median tectonic ridges near the axis of the transform valley.
Both RTIs to the north and south display asymmetric rift valleys with inside corner highs, outer corner lows, and nodal deeps at the RTIs. RTIs (e.g., see Searle and Laughton, 1977; Fox and Gallo, 1984; Tucholke and Lin, 1994). In the case of the northern RTI of the Atlantis II Transform, the relief of the inner rift valley wall is nearly double that of the outer rift valley wall.
Asymmetric rift valleys along slow-spreading centers, especially those near RTIs, have been interpreted to result from long-lived low-angle detachment faults. These detachments are thought to dip from the inner corner rift valley wall into the subsurface below the rift valley. The inner corner is envisioned as the unroofed footwall block of crustal penetrating and long-lived low-angle detachment faults (Karson and Dick, 1983; Karson et al., 1987; Karson, 1990). The outer corner is envisioned as the hanging-wall block. Detailed studies have shown that the inner corner of many rift valley walls may be largely composed of deep-seated plutonic or ultramafic rocks (e.g, Karson et al., 1987; Dick et al., 1991c; Karson and Winters, 1992; Cannat and Casey, 1995).
The SWIR lies at the extremely slow end of the spreading rate spectrum and is likely to be characterized by low magma supplies and smaller crustal thicknesses (e.g., Reid and Jackson, 1981). The inner walls of the Atlantis Transform, which represent the crust generated at the inner corners of the RTIs, have been extensively dredged, and the results show that peridotite (43%) and gabbro (24%) dominate the recovery (Dick et al., 1991a, 1991c). It has also been recognized that compositions of mid-ocean-ridge basalts (MORB) and peridotites from these dredge rocks indicate low degrees of partial melting (Johnson and Dick, 1990, 1992). These factors support the notion that the crust of the inner corners of the RTIs may be thin and may commonly represent sites where spreading was not accommodated principally by magma addition but by mechanical extension for long periods of time. Low-angle detachment faulting is regarded as the principal mechanism for accommodating extension. The northern and southern RTIs of the Atlantis Transform are likely sites in which low-angle detachment faulting and unroofing (Fig. F4) lead to exposure of deep crustal and serpentinized upper mantle rocks that are transported from the ridge axis along the transform valley and adjacent crust (Dick et al., 1991c).
There are two median tectonic ridges in the center of the transform valley (Dick et al., 1991c). In the northern half of the transform, there is a prominent median tectonic ridge that reaches 1.5 km above the transform valley floor on either side (Figs. F2, F3). It extends 110 km and plunges to the south terminating at 31º54'S. A second less prominent median tectonic ridge extends from the southern RTI to ~33º38'S. It extends ~99 km and reaches an elevation of ~250 m above the transform valley floor. The origin of median tectonic ridges have been interpreted to be associated with changes in plate motion that result in modification of the position of the principal transform deformation zone from one side of the transform valley to the other as the two RTIs are approached (Dick et al., 1991c).
Other striking features of the Atlantis Transform include two transverse ridges that lie at the top of both walls of the transform. The transverse ridges are greatly elevated with respect to the floor of the transform and the adjacent nontransform crust. Dick et al. (1991c) tentatively estimated the transverse ridges stand ~1 km above the adjacent nontransform seafloor. The ridges form prominent features that extend from the inner corner highs along the length of the active transform. Each transverse ridge consists of a series of alternating highs and lows with up to 2.5 km of relief in a longitudinal direction parallel to each transverse ridge.
The Atlantis Bank exposes gabbroic rocks that were drilled within Hole 735B during Legs 118 and 176 and in Hole 1105A of Leg 179. It is situated astride the eastern transverse ridge of the transform (Figs. F5, F6) and therefore originated at the northern RTI. Hole 735B (Robinson, Von Herzen, et al., 1989) is located 18 km west of the present-day axis of the north-south Atlantis II Transform on the Atlantis Bank. The bank is the shallowest part of the transverse ridge (Fig. F3) and is thought to represent a wave-cut platform (Dick et al., 1991c), which exposed gabbroic rocks sometime during the massif's transport to its present position along the Atlantis II Transform. This likely occurred during unroofing at the seafloor along the footwall of a long-lived normal detachment fault and formation of the rift valley inner corner high at ~11.5 Ma. It, subsequently, became part of the eastern transverse ridge (see Fig. F4). The Atlantis Bank shallows to just under 700 m, whereas the adjoining Atlantis II transform reaches a maximum depth of 6480 m. The bank is ~9 km long in a north-south direction and 4 km wide. The top of the bank is nearly flat and has proven to represent an excellent surface for both bare-rock and guide-base spudding. There was 500 m of gabbroic rock cored from Hole 735B with >86% recovery during Leg 118 (Robinson, Von Herzen, et al., 1989). In 1997, during Leg 176, Hole 735B was deepened to >1.5 km below seafloor with similarly high recovery (Dick, Natland, Miller, et al., 1999). This represents the deepest penetration of any hole drilled into an oceanic gabbroic section. The seismic velocity structure of the massif has been investigated by Muller et al. (1997) and shows that the Moho is >5 km beneath the massif. The nature of the oceanic seismic crust/mantle boundary is unknown, but the tentative velocities at the base of the crustal section appear lower than typical mantle velocities. The massif is characterized by low-relief exposures of gabbroic rocks on the top of the bank, but dredges of serpentinized peridotite have also been reported from the transform valley wall adjacent the Atlantis Bank at a water depth of ~4382 m water depth (Dick et al., 1991c). Thus, beneath the gabbroic massif of the Atlantis Bank, the Moho could represent an hydration boundary between serpentinized and unhydrated mantle assemblages. The age of the massif is ~11.5 Ma, based on isotopic dating and magnetic anomaly data over the Atlantis Bank (Stakes et al., 1991; Dick et al., 1991c).
Hammer-drilling tests were to be conducted in the general vicinity of previously drilled Hole 735B to the east of the Atlantis II Transform on the Atlantis Bank because of ideal logistical considerations. The test sites required a range of conditions from flat, bare rock outcrops to both bare rock and thinly sedimented or rubble-covered basement targets on slopes of varying inclination. This range of conditions was necessary to fully estimate the HDS capabilities in the range of operating conditions expected during hard-rock legs. The region surrounding the Atlantis Bank provides an extreme range of water depths from 700 m to more than 6 km and provides a variety of spudding surfaces ranging from relatively level massive outcroppings with clean surfaces to severely sloped, talus-covered surfaces. The first set of hammer holes were attempted directly adjacent to Hole 735B on the wave-cut platform at Sites 1104 and 1106, whereas subsequent hammer drilling was to occur on the slopes adjacent to the platform.
Results from Hole 735B have shown that drilling deep into the gabbroic sections of the oceanic crust can be accomplished under the right conditions. Previous drilling in the region during Leg 118 (Natland et al., 1991; Dick et al., 1991a) has shown that the core consists of foliated metagabbro near the surface (Unit I, 0-37.41 mbsf), olivine gabbro (Unit II, 37.41-170.22 mbsf), followed downward by olivine gabbro with disseminated oxide gabbro (Unit III, 170.22-223.57 mbsf), massive oxide gabbros (Unit IV, 223.57-274.06 mbsf), massive olivine gabbro (Unit V, 274.06-382.40 mbsf), and olivine gabbro, oxide gabbro and troctolites (Unit VI, 382.40-500.70). The deeper penetration of Leg 176 to ~1500 m depth (Dick, Natland, Miller, et al., 1999) will add to this stratigraphic succession, but the shallowest part of the section will be of particular interest during Leg 179 because it may cover part the same stratigraphic interval as that anticipated in coring Hole 1105A. The most distinctive unit in Hole 735B is Unit IV, which consists of massive oxide gabbros. Unit IV also shows an elevated magnetic susceptibility signature when compared with the units above and below (Pariso et al., 1991).
The bulk rock and mineral geochemistry in the upper 500 m of the section (Bloomer et al., 1991; Natland et al., 1991; Dick et al., 1991a; Hebert et al., 1991; Ozawa et al., 1991) shows that the section has a wide range of geochemical signatures from very primitive (bulk rock Mg# of 0.80-0.88) to very fractionated (bulk rock Mg# of 0.23-0.30). In general, the bulk and mineral chemistries of the samples clearly show two groupings that broadly divide the oxide gabbro or gabbronorite and the olivine gabbros. Olivine gabbros show primitive bulk rock and mineral compositions, whereas oxide gabbros tend to show more fractionated mineral chemistry. Both groupings overlap in a downhole plot but are clearly distinctive trends. Both Hebert et al. (1991) and Ozawa et al. (1991) show that the olivine gabbro samples can be divided into two units downhole. One from 0-270 m and the other from 270-500 m. Each unit shows a trend from somewhat more primitive mineral chemistries at the base to more fractionated upward. This type of cryptic chemical variation in the olivine gabbros is what would be expected of cyclic units caused by fresh influxes of primitive magma followed by fractionation (e.g., Irvine, 1979; Komor et al., 1985). The reset at 270 m to a higher Mg# could be interpreted to mark a major fresh input of primitive magma. Superimposed on this major reset are finer scale cyclic cryptic chemical variations (Hebert et al., 1991), which are not as well defined because of the scale of sampling. Each reset generally yields higher Mg# in clinopyroxene and olivine and higher An content of plagioclase.
The oxide gabbros and gabbronorites over the same stratigraphic intervals show the opposite trend. Instead of becoming more fractionated upward in each cycle, the well-defined upper trend becomes more primitive from 270 mbsf upward (Ozawa et al., 1991; Hebert et al., 1991). The lower unit, although more poorly defined, also becomes more primitive upward. This provides an overall downhole trend, which is an inverse olivine gabbro trend. The oxide gabbros, although abundant higher in the section, also become scarcer lower in the hole. Dick et al. (1991a) interpret the oxide gabbros and more fractionated rocks to represent the products of evolved intercumulus melts. Hebert et al. (1991), however, emphasize that there is a continuum of mineral compositions between oxide gabbro and olivine gabbro. The bimodal distribution of rock types and mineral chemistries over the same stratigraphic interval, however, is somewhat unique.
Magmatic, crystal-plastic, and brittle structures have been documented throughout the core. Most of the core intervals are free of significant crystal-plastic deformation and may or may not be characterized by weak magmatic foliations and fabrics (Cannat, 1991). There are many inclined ductile shear zones with normal shear sense (Cannat, 1991). These appear more prevalent near the top of the hole. Cataclastic zones and brittle faults have also been noted. Of particular note is the strong correlation between zones of crystal-plastic deformation and oxide-rich gabbros (Natland et al., 1991; Dick et al., 1991c; Cannat et al., 1991).
Metamorphism in core from Hole 735B ranged from static to dynamic (Stakes et al., 1991; Kempton et al., 1991; Alt and Anderson, 1991). Dynamically metamorphosed rocks generally showed the highest temperature alteration and include granulites with largely anhydrous assemblages to amphibolites with brown to green amphibole. Amphibolite gneisses were documented at the top of the hole. Statically metamorphosed rocks show high (>600ºC) to moderate temperature (440º-600ºC) alteration, the latter usually associated with fractures filled by vein material. Low-temperature smectite, zeolite, calcite, and prehnite were noted along fractures and represent latest alteration. Oxygen and Sr isotope data generally suggest low water-rock ratios during alteration (Kempton et al., 1991; Stakes et al., 1991).
Paleomagnetic studies of the core indicate that the core is reversely magnetized and that samples have consistently undergone an ~20º rotation because the magnetic inclination is steeper than would be expected for the region (Pariso et al., 1991). In addition, the presence of oxide-rich gabbro and its magnetic signature suggests that the gabbros may represent an important source component for marine magnetic anomalies (Pariso et al., 1991; Dick et al., 1991c).
The selection of a location for Site 1105 was aided by a site survey cruise of the James Clark Ross, which was conducting bathymetric, magnetic, and sampling surveys of the Atlantis Bank as the JOIDES Resolution arrived. Data transfer between the James Clark Ross and the JOIDES Resolution allowed an adequate site to be chosen for drilling, which was a considerable distance from Hole 735B. The site had to meet certain requirements because we chose not to employ a hard-rock guide base to achieve maximum penetration in the short time available. The site was at the crest of the Atlantis Bank where the surface relief was minimal. We achieved a bare-rock spud by using a tricone bit and drilled to ~15 mbsf. A modified free-fall funnel was installed, and drilling into gabbroic rock commenced.
The site is close to a ridge-axial trend with respect to Hole 735B but more distal from the north-south Atlantis II Transform, which lies directly to the west of Hole 735B. The site was chosen to avoid a duplication of Hole 735B core by drilling at proximal Site 1104, where the hammer tests were conducted for the first 3 days. At the same time, we tried to utilize Hole 735B as a reference section to attempt to accomplish the following objectives:
If successful, the correlation experiment could yield a minimum measure of the dimensions of subaxial magma chamber systems and continuity of structure and metamorphic processes along the strike of the ridge axis at very slow-spreading centers. If correlations are unsuccessful, the scale of the experiment will place limits on the dimensions of igneous units, the former size of continuous magma chambers, and the continuity of structures. Correlation will be attempted on the basis of detailed and integrated data sets including core descriptions and subsequent shore-based laboratory analyses to establish cryptic chemical and mineralogical variations, as well as alteration and structural profiles of the core. Data from a full logging program will also be used in the correlation attempts.
The axial thermal structure (Kuznir and Bott, 1976) along mid-ocean ridges is likely to vary significantly as a function of spreading rates and proximity to hot spots. Based on the ability to image crustal level magma chambers using seismic methods, fast-spreading centers are usually considered to maintain small, but likely steady-state magma chambers (e.g., Detrick et al., 1987; Harding et al., 1990; Sinton and Detrick, 1992), whereas slow spreading center data has not yet suggested any semblance of steady-state chambers (Fowler, 1976; Fowler, 1978; Nisbet and Fowler, 1978; Detrick et al., 1990). It should be noted that the seismic results used at fast-spreading centers having lower bathymetric relief are not directly comparable in resolving power to those at high relief, slow-spreading mid-ocean ridges. In addition, refraction studies along slow-spreading ridges have resolved lower velocity crust along the rift valley (Purdy and Detrick, 1986). Overall, however, the data indicate that the plutonic foundations of slow-spreading crust are likely to be created by solidification of smaller ephemeral magma chambers, the size and dimensions of which we have little direct information (Fig. F7). Most investigators place the maximum size of slow-spreading magma chambers at <1-2 km (e.g., Nisbet and Fowler, 1978; Detrick et al., 1987).
Drilling into tectonic windows of the plutonic portion of the ocean crust with a dense closely spaced array of holes that cover a substantial depth range is likely the only viable method in attempting to map and firmly establish the dimensions and nature of transient magma chambers at slow-spreading centers. Where several holes have been drilled into tectonic windows (e.g., MARK and Hess Deep), the shallow depths of penetration, the recovery rates, and/or lateral distances between holes were generally not conducive to lateral correlation of sections. The only hole to date that has achieved significant penetration into the plutonic crust, as well as an exceptionally high recovery rate (86%), is Hole 735B along the SWIR, which was recently deepened and achieved penetration depths of 1508 mbsf during Leg 176 (Dick, Natland, Miller, et al., 1999). The core from Hole 735B provides the first almost-continuous section of oceanic gabbros sampled by ocean drilling. During Leg 179, we had the opportunity to drill and core a hole at a moderate distance of 1.2 km from Hole 735B in an attempt to laterally correlate units between holes. Whereas the depth of penetration was modest (158 mbsf) compared to that of Hole 735B, the recovery rate was nearly identical.
Ophiolite sections are generally regarded as representative of obducted segments of oceanic crust derived from or incorporated into a forearc setting before their obduction onto continental crustal blocks (Dewey, 1976; Casey and Dewey, 1984). Their exact origin is problematic because of trace element signatures in volcanics that generally indicate an arc tectonic environment, although the same signature has recently been observed along a mid-ocean ridge (the Chile Rise) by Klein and Karsten (1995). Without addressing these uncertainties, it is sufficient to say that it is likely that classic ophiolite sequences, including pillow lavas, sheeted dikes, gabbroic rocks, and ultramafic tectonites, represent some type of ocean crust and mantle formed in extensional environments that may have similarities to mid-ocean ridges. Thus, analogies with the plutonic sections of ophiolites may indeed help in deciphering the nature of the plutonic foundations of oceanic crust, especially when our view is limited to the scale of a borehole core.
The overall stratigraphy of ophiolite plutonic sections is very well summarized from base to top by many studies (e.g., Coleman, 1977; Dewey and Kidd, 1977; Casey et al., 1981). The plutonic stratigraphy of ophiolites is generally defined by the composition of the cumulates and the presence or absence of fine-scale igneous layering (centimeter to meter scale). Generally, at the base it consists of what is thought to represent an ultramafic cumulate section directly above depleted mantle residual harzburgite or less depleted lherzolite. This is followed upward by a strongly interlayered mafic and ultramafic transition zone, followed by a layered gabbro section, an isotropic gabbro section, and commonly a dike-gabbro transition (see Fig. F7). These definitions are somewhat simplified and at times misleading. The most strongly layered rocks (centimeter to meter scales) are at the base of the plutonic section, which consists of ultramafic and transition zone cumulates that are commonly tectonized, and lowermost part of the layered gabbro section. In general, as one progresses upward through the layered gabbro section, the intervals of monotonous nonlayered gabbros between finely (centimeter to meter scale) interlayered gabbroic rocks increase until strongly layered gabbros disappear. The zone of nonlayered gabbros at the top of the plutonic section is called the isotropic gabbros, but they are hardly isotropic. They are simply nonlayered but commonly contain faint banding and significant variations in grain size and modal mineralogy. They are also characterized by the highest oxide abundances and contain intrusive bodies of trondhjemite and other felsic rocks, as well as scarce to abundant crosscutting diabase dikes.
In sections below the isotropic gabbro, which show layering, lateral continuity of the fine-scale igneous layering is generally lacking, and the layers tend to pinch out over lateral distances equivalent to outcrop scales. This is unlike continental layered intrusions where, in general, lateral continuity of layering on a kilometer scale is well demonstrated and where, rarely, a single layer can be followed for more than 25 km (e.g., Irvine, 1979). These types of scales of lateral continuity are never approached in the case in ophiolite plutonic sections. The fine-scale (centimeter to meter scale) layers in ophiolites cannot be followed laterally for more than ~100-200 m and commonly much less because they pinch out. There are larger scales of layering, as much as 40 to 200 m, that can be traced laterally for kilometers and can be correlated from one section to the next (e.g., Casey et al., 1981; Casey and Karson, 1981; Komor et al., 1985); however, the fine-scale layering on the scale observed in oceanic gabbroic core is generally the laterally discontinuous type within each distinctive unit. The fine-scale layers are also characterized by variable orientations in ophiolites with inclinations that range from paleohorizontal to paleovertical (Dewey and Kidd, 1977; Casey and Karson, 1981; Smewing, 1981; Casey et al., 1983). Similar dipping layers have been observed in ODP cores (e.g., Cannat, Karson, Miller, et al., 1995).
The large-scale units defined in ophiolites generally consist of interlayered intervals with distinctive rock types, such as troctolite, pyroxenite, or oxide gabbro. In addition, the broad-scale map unit contacts between isotropic gabbro, layered gabbro, transition zone, and ultramafic cumulates can be traced laterally. These units, however, typically show extreme variation in thickness. (Casey et al., 1981). For example, the ratio between layered gabbros and isotropic gabbros in a crustal section can change significantly along strike by up to a factor of 3 or 4. Over a scale of several kilometers, the isotropic gabbro section can be the dominant part of the gabbroic interval or can be almost nonexistent.
In summary, the lack of lateral continuity of fine-scale layering in the oceanic crust, but improved continuity for larger scale units, should be expected in the oceanic crust. Another characteristic of ophiolites is the lateral heterogeneity of broad-scale unit thicknesses. Likewise, this may be a typical feature in the oceanic crust.
How the ophiolite model (Fig. F7) of the plutonic section fits with the pseudostratigraphy predicted by small ephemeral chambers, which might exist at slow-spreading magma-starved ridge segments, is not exactly known. Primitive mantle-derived MORB magmas forming discrete magma chambers in the crust should generate the same type of cumulates observed in ophiolites or layered intrusions but within each small intrusion. The crystallization sequence of MORB magmas should generate dunite cumulates, followed by troctolites, olivine gabbros, gabbros, oxide gabbros or gabbronorites, and, finally, felsic rocks within each small magma body as it fractionates from primitive MORB (Fig. F7). Alternatively, magmas fractionate olivine in the colder upper lithospheric mantle beneath slow-spreading centers and arrive at crustal levels after fractionating enough olivine to only generate gabbroic rocks at crustal levels (Cannat and Casey, 1995; Niu, 1997; Cannat et al., 1997; Casey, 1997; Niida, 1997). This, however, is inconsistent with the fact that picritic basalts with only olivine and spinel on the liquidus are known to pass through the subaxial plumbing system and erupt along some mid-ocean ridges (Melson and O'Hearn, 1986; Natland, 1990; Perfit et al., 1996). Small ephemeral chambers beneath magma-starved slow-spreading centers like the SWIR might predict a complex stratigraphy of mixed ultramafic, gabbroic, and felsic rocks solidified in each small chamber repeating itself within a stratigraphic section cored through the crust. They also should show many crosscutting igneous intrusions and relationships in order to feed magma chambers at different levels in the crust, sheeted dikes, and pillow lavas above. This model is similar to the sill model proposed for the Oman ophiolite (Boudier et al., 1996), but the model predicts that the lower crust should be riddled with vertical feeder dikes, as well as sill systems, to feed multiple chambers in the plumbing system (Fig. F7). Sinton and Detrick (1992) propose that the subaxial regions of slow-spreading centers represent crystal mushes with variable amounts of melt porosity, through which magma is delivered from the mantle to the surface. This mechanism would tend to reduce lithologic and chemical variability within the mush zone because melt delivered by porous flow will continually react with the crystalline material through which it passes.
Drilling during Legs 118 and 176 provides the most extensive drill core yet recovered from a plutonic section generated at a slow-spreading center. Drilling during Leg 179 at Site 1105 provided the first samples and logging information of an offset hole necessary to attempt to laterally correlate sections of the plutonic crust for distances >1 km. Although Hole 1105A is more limited in penetration, it provides internal checks that help constrain the results from each of the holes drilled on the Atlantis Bank.
Investigations carried out in the region of the Atlantis Bank along the Atlantis II transform have been extensive and represent the location of the deepest penetration yet by ODP into the plutonic foundations of oceanic crust. This site provides an ideal location to test a new engineering approach to bare rock drilling in hard-rock terrane on the ocean floor. The investigations carried out during Leg 179 will attempt to improve the ability of the drillship to spud and create stable holes in hard-rock environments with a new hammer drill-in casing system. In addition, contingency rotary coring in the region of the Atlantis Bank will attempt to refine models of the architecture of the plutonic foundations of the ocean crust by drilling in gabbroic rock offset by 1.3 km from reference Hole 735B.
1Examples of how to reference the whole or part of this volume can be found under "Citation" in the preliminary pages of the volume.