What Lies Beneath: The Formation and Evolution of Oceanic Lithosphere

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. What Lies Beneath: The Formation and Evolution of Oceanic Lithosphere Katsuyoshi Michibayashi, Masako Tominaga, Benoit Ildefonse, Damon Teagle


INTRODUCTION
Scientific ocean drilling commenced through the initiation of Project Mohole in 1961, about the same time as Apollo moon landing ambitions were first artic ulated. It has been almost 60 years since the American Miscellaneous Society con ceived of the idea for Project Mohole and 50 years since the launch of the Deep Sea Drilling Project (DSDP) in 1968. Scientific ocean drilling is an essential approach to directly access Earth's interior and is arguably science's most successful international collaboration. Although this cooperation has greatly expanded from the DSDP (1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983), through the Ocean Drilling Program (ODP, 1983(ODP, -2003 to the Integrated Ocean Drilling Program (IODP, 2003(IODP, -2013, and to the International Ocean Discovery Program (IODP, 2013(IODP, -2023, gaining a better understanding of the dynamics of our planet remain challenging due to the technical difficulties of drilling holes deeper than 100-200 m into the oceanic crust's igneous basement. A compilation of holes drilled into in situ oceanic crust by scientific ocean drill ing since the beginning of DSDP through 2018 highlights the problem: only 38 holes are deeper than 100 m and only 20 are deeper than 200 m (Figures 1 and 2; e.g., Ildefonse et al., 2007bIldefonse et al., , 2014. The first attempt was DSDP Hole 332A drilled on Leg 37 in 1974 (Aumento et al., 1977). The total amount of igneous oceanic crust recovered represents less than 2% of the material archived in the DSDP, ODP, and IODP core repositories. Despite this rel ative paucity of material, scientific ocean drilling has provided essential and hith erto unavailable observations that are advancing our understanding of the pro cesses that "repave" nearly 70% of Earth's surface over short geological time scales (<200 million years). We have bet ter knowledge of oceanic crust architec ture, magmatic accretion processes in ABSTRACT. Sampling the upper mantle via scientific ocean drilling remains elusive. Although the technologies required for drilling to the Moho still don't exist, we have made significant progress over the last five decades in piecing together the complex geology of the oceanic crust. Here, we highlight key findings that reveal the architecture of oceanic crust and the thermal, physical, and chemical processes that are responsi ble for the growth and structure of the oceanic lithosphere. These advances result from enduring efforts to drill and collect downhole geophysical logs of oceanic crust near both slow and fast spreading ridges.

FIGURE 1.
Compilation showing holes drilled >100 meters below seafloor (mbsf) into the basement of intact oceanic crust and tectonically exposed lower crust and upper mantle from 1968 to 2018 (see drill hole sections in Figure 2). Sites mentioned in the text are labeled. Seafloor age based on age grid by Müller et al. (2008, revised version 3; www.earthbyte.org/). This map does not include "hard rock" drill holes in seamounts, oceanic plateaus, back-arc basement, hydrothermal mounds, or passive continental margins. the centers of midocean ridge spread ing centers, and the nature and magni tudes of hydrothermal exchange between the ocean and the oceanic lithosphere, and scientific ocean drilling samples led to the discovery of a deep microbial rock hosted biosphere.
With the results from 50 years of scien tific ocean drilling, we now know that in all ocean basins a volcanic basement lies beneath an almost omnipresent blanket of sediments, created by a system of mid ocean ridges that together form the larg est magmatic province on Earth, gener ating more than 20 km 3 of new oceanic crust each year. Roughly twothirds of the magma derived from the partial melt ing of upper mantle peridotite cools and crystallizes as plutons in the lower por tion of the oceanic crust; the remainder erupts as basalt and forms the upper one third of this basement.
Here, we focus on the importance of basement drilling and the advance ments in our understanding of the key differences in ocean crust architecture as a function of plate tectonic setting and related thermal, physical, and chemical processes. We summarize early attempts in the 1960s and current plans to reach the Mohorovičić Discontinuity (Moho) at the lower ocean crust boundary with the upper mantle, and we discuss how scien tific ocean drilling has informed us on the major differences in oceanic crust created in fast and ultraslow spreading settings.

PROJECT MOHOLE
Project Mohole 1 has been an iconic aspi ration in the Earth sciences, a fundamen tal driver of scientific ocean drilling and a focus of five decades of enduring col laborations between the United States and its international partners (Hsü, 1992). Originally, Project Mohole pro vided a geoscience foil to the nascent US space program. The essence of Project Mohole was to retrieve samples of Earth's mantle by penetrating the Moho, a major global seismic boundary between Earth's crust and upper mantle. Seismologists had already subdivided the oceanic crust into seismic layers: Layer 1 comprising low Pwave velocity sediments (V p < 3 km s -1 ); Layer 2 having low Pwave velocity and a steep velocity gradient, with V p ranging from ~3.5 km s -1 to ~6.7 km s -1 , typical of basalt; and Layer 3 having high velocity and a more gentle velocity gradient (V p of 6.7-7.1 km s -1 ) that we now know is typi cal of gabbro. However, an abrupt increase at the base of Layer 3 to V p > 8 km s -1 was found to mark the Moho and was inter preted to be the boundary between the gabbros of the lower oceanic crust and the ultramafic (peridotitic) rocks of the uppermost mantle.
The ultimate proposal was to drill to the Moho in the deep ocean where Earth's crust is relatively thin (~6 km;National Research Council, 1957;Bascom, 1961). Attempting such an effort on land would have been impractical because the drill ing equipment would have to withstand high in situ temperatures at great depths while drilling through the much thicker (>30 km) continental crust. In addition, cores sampled by ocean drilling offer a simpler and "cleaner" record of major geological processes, rather than the com plex geology sampled by a terrestrial deep hole that would have resulted from mul tiple global tectonic ~400-500 million yearlong "Wilson cycles. " If success ful, the highly ambitious and technically challenging Project Mohole would have yielded new observations on the age and composition of the seafloor, while provid ing evidence for the theory of continental drift that at the time remained controver sial and strongly debated.
Project Mohole comprised a three phase plan (National Research Council, 1959

FIGURE 2.
Compilation showing scientific ocean drilling holes that penetrated >100 m into the basement of intact oceanic crust and tectonically exposed lower crust and upper mantle from 1968 to 2018 (see drill hole locations in Figure 1). The number designation for each hole and the recovery (in percent) for each of the basement lithologies are indicated. This compilation does not include "hard rock" drill holes in seamounts, oceanic plateaus, back-arc basement, hydrothermal mounds, or passive continental margins.
drilling vessel for deepwater operations and testing the vessel and equipment in deep water far offshore. This required the development of new capabilities, includ ing: (1) navigational and thruster tech nologies to keep a floating vessel at a sin gle deepocean location, now known as "dynamic positioning" and a univer sal feature on any modernday research vessel, and (2) a strategy that would allow subsequent visits to reenter the drill holes and resume drilling efforts (Bascom, 1961 , 1961). This was the first in situ demonstration that the oce anic basement comprises (young) basal tic lavas, and that seismic Layer 2 is basalt. Project Mohole Phase 1 was a major early step in the exploration of Earth's inte rior, with scientists receiving a congratu latory telegram from US President John F. Kennedy: "The success of the drill ing in almost 12,000 feet of water near Guadalupe and the penetration of the oceanic crust down to the volcanic forma tions constitute a remarkable achievement and an historic landmark in our scientific and engineering progress" (The National Academies of Sciences, Engineering and Medicine, 2011; see Becker et al., 2019, in this issue, Figures 1 and 2 for photos of CUSS I and this telegram).
Notwithstanding this early success, Project Mohole became mired in political controversy and was terminated in 1966 before further holes were drilled. Despite Project Mohole not achieving its original goal of drilling to the mantle, the project contributed to a "movement" in the solid Earth community, cumulating in the global acceptance of the theory of plate tectonics. Moreover, Project Mohole not only showed that scientific ocean drilling could successfully drill into and recover core samples from oceanic basement but also illustrated that ocean drilling is an essential tool for gathering other wise inaccessible information about how our dynamic planet operates (Teagle and Ildefonse, 2011). Project Mohole led to formation of the US Deep Sea Drilling Project (DSDP), whose first expedition sailed in 1968.

EARLY YEARS: PENROSE MODEL AND CORING IN OCEANIC CRUST
The earliest years of DSDP concentrated on recovering long sediment cores to refine marine sedimentbased biostra tigraphy models and to validate the the ory of seafloor spreading by dating the sediments directly overlying the oceanic basement (DSDP Leg 3; Maxwell et al., 1970). When the very top of the oceanic basement was "tapped" below the sed iment column, recovered samples were recognized as pillow lavas, providing the first direct evidence of lava that was rapidly cooled in a subaqueous environ ment. These samples became the center of a debate on the origin of commonly juxtaposed rock strata observed in loca tions such as the Troodos Massif, Cyprus (e.g., Gass, 1968;cf. Miyashiro, 1973) and other orogenic belts. Geologists work ing on these socalled ophiolites reached a consensus statement during the 1972 Penrose Field Conference that defined these rock sequences in the context of the new paradigm of seafloor spreading, in what is now referred to as the Penrose model (Anonymous, 1972). This state ment developed the widely accepted model that ophiolites are ancient and largely intact sections of oceanic crust preserved on land that comprise, from bottom to the top: (1) ultramafic rocks of the upper mantle, (2) gabbros, (3) a sheeted dike complex, (4) basal tic lavas, commonly pillow basalts, and (5) associated sedimentary deposits such as ribbon cherts, thin shale inter beds, and minor limestones ( Figure 3a). The Penrose model raised the enduring science question as to whether ophio lites represent a direct analog for in situ oceanic crust beneath the modern sea floor (e.g., Panayotou, 1980;Gass, 1990), and this question, in turn, has been an important motivation for drilling the oceanic crust (e.g., Dilek et al., 2000).
The first international efforts to drill deeply into the oceanic crust were DSDP Leg 34 in 1973-1974 on the Nazca Plate in the Eastern Pacific Ocean (Yeasts et al., 1976), and DSDP Leg 37 in 1974 on the western flank of the MidAtlantic Ridge, south of the Azores Plateau (Aumento et al., 1977). These legs recovered, for the first time, tens of meters of basaltic core samples from upper oceanic crust (e.g., 59 m in Site 319 during Leg 34; >100 m in Holes 332A,B and 333A during Leg 37; Figures 1 and 2). It is also note worthy that cores from Leg 37 Site 334 in the Atlantic recovered small amounts of gabbro and serpentinized peridotite from the presumed deeper layer in a typ ical Penrose style of oceanic lithosphere, at relatively shallow (117 meters below sedimentbasement contact) subseafloor depths (Aumento et al., 1977), suggest ing a vertical and lateral crustal heteroge neity and demonstrating that the Penrose model is an end member model itself ( Figure 3; Ildefonse et al., 2014).

DEEP DRILLING IN FAST-SPREADING CRUST
Although less than 20% of the mod ern midocean ridge system is creating new seafloor at fast rates (>80 mm yr -1 full rate), nearly half of the oceanic crust created over the last 200 million years formed at fastspreading ridges (Teagle et al., 2012;Ildefonse et al., 2014). Deep drilling into oceanic crust at a few sites in the Eastern Pacific, including the Cocos Plate (ODP Holes 504B and 1256) and the Hess Deep (IODP Site U1415), has led to a widely accepted model of ocean crust architecture that is very similar to, and confirms in large terms, the Penrose model. Drilling has also provided insights into the nature of key seismic boundaries found in fastspreading oceanic crust and into the role of alteration, grain size and texture, and composition in controlling these boundaries.

DSDP/ODP Reference Hole 504B: Nazca Plate
DSDP/ODP Hole 504B, located in 6 mil lion year old crust 200 km south of the Costa Rica Rift in the eastern equato rial Pacific, has long been a "reference" site for intact oceanic crust formed at an intermediate to fastspreading cen ter (Figures 1 and 2) between the oceanic Cocos (north) and Nazca (south) tec tonic plates. It is the deepest hole drilled into the igneous oceanic crust, penetrat ing 2,111 meters below seafloor (mbsf) and 1,836.5 m into the subbasement over the course of seven ODP and DSDP legs since 1979 (DSDP Legs 69, 70, and 83, and ODP Legs 111, 137, 149, and 148;Cann et al., 1983;Anderson et al., 1985;Alt et al., 1986Alt et al., , 1993Alt et al., , 1996Becker et al., 1988Becker et al., , 1992Dick et al., 1992). The hole was also visited during DSDP Leg 92 in 1983 for downhole logging and sampling of bore hole fluids (Leinen et al., 1986) and will be revisited in 2019 (IODP Expedition 385T; Tominaga et al., in press). ODP Leg 148 in 1993 was the last time Hole 504B was deepened, this time by 111 m. Further penetration is currently prevented because portions of a drill bit are stuck in the hole (Alt et al., 1993).
The lithologic sequence in Hole 504B consists (from top to bottom) of 274.5 m of sediment, 571.5 m of volcanic rocks, a 209 m transition zone, and 1,050 m of a sheeted dike complex (Figure 2; Alt et al., 1996). The hydrothermal alteration of the volcanic section in Hole 504B involves a series of processes that entail interaction with oxidizing seawater at low tempera tures, with intensity decreasing down ward. These processes and their effects on the volcanic section are generally sim ilar to those in other oceanic upper crustal sections. The transition zone and upper dikes (down to 1,500 mbsf) were altered in a subsurface mixing zone, where hydro thermal fluids upwelling through the dikes mixed with cooler seawater circulating in the overlying more permeable volcanic rocks. Mineral assemblages in the cored permeable pillow basalts in the transi tion zone indicate that during hydrother mal circulation, a maximum temperature of ~350°-380°C may have been reached. This is typical of greenschist facies meta morphism that includes such alteration minerals as chlorite, actinolite, and albite oligoclase (Alt et al., 1996). The lower dikes (1,500-2,111 mbsf) were hydrother mally altered at temperatures exceeding 400°C, resulting in the formation of horn blende and calcic secondary plagioclase, which then subsequently were overwrit ten by similar reactions that produced the pillow basalt greenschist assemblages at ~300°-400°C. Alteration of the sheeted dikes from Hole 504B is heterogeneous, with recrystallization controlled by frac turing and fluid access (Alt et al., 1996). Defining the position of the seismic transi tion between Layer 2 (basalts) and Layer 3 (gabbros) in Hole 504B depends upon the scale of observation, but appears to cor relate with observed progressive changes in porosity and hydrothermal alteration (Alt et al., 1996). Therefore, the nature of the transition from sheeted dikes to gab bros in Hole 504B remains obscured.

ODP-IODP Superfast Hole 1256D: Cocos Plate
ODP Hole 1256D (Figures 1 and 2 Boudier and Nicholas (1985) Penrose Model as modi ed by Dick (1989) Cannat  Wilson, 1996) at the East Pacific Rise. The site formed on a ridge segment that is at least 400 km long, located ~100 km north of the ridgeridgeridge (RRR) tri ple junction between the Cocos, Pacific, and Nazca Plates. The deep drilling campaign at Site 1256 was aimed at understanding the formation, architecture, and evolution of oceanic crust formed at "superfast" plate spreading rates. It has been the focus of four scientific ocean drilling cruises (ODP Leg 206 and IODP Expeditions 309, 312, and 335; Wilson et al., 2003;Teagle et al., 2006Teagle et al., , 2012. Hole 1256D was the first scientific ocean drilling bore hole prepared for deep drilling in oceanic crust. A large funnel, or reentry cone, was installed at the seafloor and then secured downhole with almost 270 m of 16inch casing through the 250 m thick sedimen tary overburden, and then cemented into the uppermost basement (Wilson et al., 2003). During ODP Leg 206, the bore hole was deepened through an ~810 m thick sequence of basaltic lavas and a thin (~346 m) sheeted dike complex, the lower 60 m of which shows evidence for the formation of granoblastic textures (i.e., rocks with a dense arrangement of large equidimensional minerals with sutured boundaries) that typically result from high temperature contact meta morphism . During IODP Expedition 312, the first gabbroic rocks were encountered at 1,407 mbsf Teagle et al., 2006) at a depth where the hole entered a com plex dike-gabbro transition zone that includes two gabbro lenses (20-50m thick) intruding into basalt dikes with the same hightemperature granoblastic textures (Figure 4). IODP Expedition 335 returned to Hole 1256D in 2011 with the ambition of deepening the hole several hundred meters into the cumulate gab broic rocks of intact lower oceanic crust. However, drilling in this hole advanced only minimally to 1,521 mbsf (Figure 4), as a number of significant engineering challenges were encountered during the expedition that prevented deepening of the hole beyond this "hardened" meta morphic unit (Teagle et al., 2012). Based on regional seismic refraction data, the transition from basalt Layer 2 to gabbro Layer 3 at Site 1256 occurs between 1,200 m and 1,500 m into base ment (Wilson et al., 2003). An examina tion of shipboard and postcruise discrete sample measurements, wireline logging data, and vertical seismic velocity profil ing suggests that the base of Hole 1256D is at, or very close to, the Layer 2-3 tran sition (Swift et al., 2008;Gilbert and Salisbury, 2011 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 have lost more than 30% of its original liquid mass, implying that at least 300 m of cumulate gabbro formed as a residue during ocean crust formation must be present in the crust below the present base of Hole 1256D . However, encountering gabbro already at a shallower depth within Layer 2 rein forces previous inferences that factors such as porosity and hydrothermal alter ation (Detrick et al., 1994;Alt et al., 1996;Carlson, 2010) are more important than rock type or grain size in controlling the location of the seismic Layer 2-3 transi tion. This is an important advance in our understanding of oceanic crustal archi tecture, despite the fact that the Moho at the base of the oceanic crust could still be thousands of meters below the hole. Future scientific ocean drilling and the deepening of Hole 1256D is required to characterize the true nature of the Layer 2-3 "basalt to gabbro" seismic tran sition at Site 1256.

IODP Site U1415: Hess Deep
IODP Hess Deep Expedition 345 in 2012/2013 was designed to sample lower crustal primitive gabbroic rocks that formed at the fastspreading East Pacific Rise (EPR) in order to test models of magmatic accretion and the intensity of hydrothermal cooling at depth (Gillis et al., 2014a(Gillis et al., , 2014b (Gillis et al., 1993). Previous studies of known seafloor exposures of lower plu tonic rocks have suggested that layering exists in the gabbroic section. IODP Site U1415 recovered primi tive olivine gabbros and troctolites (a pyroxenedepleted version of gabbro) at one 35 m deep hole (U1415I) and two ~110 m deep holes (U1415J and U1415P shown in Figures 1 and 2) located within 100 m of each other (Gilles et al., 2014b). The cores recovered at Site U1415 can be placed more than 2 km beneath the sheeted dikeplutonic transition and thus may represent the lower plutonic half of the EPR fastspreading crust (Gilles et al., 2014a). The abundance of layering in the material recovered from Site U1415, along with the absence of other inter mixed, more evolved lithologies, dis tinguishes the lower gabbroic crust at Hess Deep from crustal sections recov ered from other ODPIODP expeditions to slowspreading ridges. These obser vations support previous models that invoke strong spreading rate and thermal control on magma chamber processes at midocean ridges; however, the style of layering and banding, as well as the observed lithologies, differ from the mid ocean ridge basaltlike Oman ophiolite, which has been used as a fast spreading ridge analogue and informed the ini tial Penrose model (Figure 3; Gillis et al., 2014a). IODP Hess Deep Expedition 345 thus provides a reference section for primitive fastspreading lower crust that did not previously exist. This highlights the need for scientific ocean drilling to address questions related to the origin, evolution, and heterogeneity of the lower oceanic crust.

DEEP DRILLING IN SLOW-SPREAD CRUST AND OCEANIC CORE COMPLEXES
It is well known from dredging and ROV sampling that a continuous gabbroic layer does not exist at slowspreading ridges and at tectonically formed oceanic core complexes exposed in these slow spread ing environments (e.g., Whitehead et al., 1984;Mutter et al., 1985;McCarthy et al., 1988;Dick, 1989;Cannat, 1993;Tucholke and Lin, 1994). Moreover, the abundance of serpentinized peridotite in dredge hauls from rift valley and fracture zone walls (Aumento and Loubat, 1971;Thompson and Melson, 1972;Fisher et al., 1986;Dick, 1989;Cannat, 1993) raised the possibility that serpentiniza tion can be a significant component of seismic Layer 3 "gabbros" in these set tings (Figure 3), as originally suggested by Hess (1962). Without scientific ocean drilling, no truly representative section of seismic Layer 3 (which may not be the same everywhere) is likely to be obtained in situ in these oceanic slow spread ing settings and core complexes, leaving its composition, state of alteration, and internal structure almost entirely a mat ter of inference.

ODP Hole 735B: Atlantis II Fracture Zone
In 1997, ODP drilled Hole 735B through a 1,508 m section of coarse gabbro in tectonically exposed lower crust on a wavecut platform that flanks the Atlantis II Fracture Zone on the slowspreading Southwest Indian Ridge (Figures 1 and 2). The sequence of rocks sampled in Hole 735B ( Figure 5) is unlike that in a Penrosetype ophiolite, in Hess Deep, or in layered intrusions found on land. Some of its attributes, including the lack of welldeveloped layering, and the presence of small 100 m to 500 m intru sions, are similar to the typical structural characteristics of ophiolites believed to have formed in slowspreading environ ments, such as the Trinity or Josephine ophiolites, although these onland ophi olite sequences are believed to be incom plete (Dick et al., 1999). The results from Hole 735B documented a systematic vari ation in igneous petrology, structure, and alteration with depth, unlike that expected in crust formed in association with large magma chambers or even melt lenses now inferred to exist beneath fastspreading ridges (Dick et al., 1999). They provide a first assessment of synkinematic igneous differentiation in which the upper levels of the gabbroic crust are enriched in late differentiated melts by means of tectonic processes, rather than the simple gravita tionally driven crystallization differenti ation often seen in layered intrusions of large terrestrial magma chambers.

ODP Legs 109 and 209: Mid-Atlantic Ridge Rift Valleys
ODP Leg 109, Site 670, on the west wall of the MidAtlantic Ridge median valley near 23°10'N targeted the lowermost oce anic crust, and for the first time drilled and sampled serpentinized mantle peri dotites (Bryan et al., 1988). In the same area, south of the Kane Fracture Zone, a total of 95 m of serpentinized peridotites were recovered from a 200 m deep hole at Site 920, ODP Leg 153 (Figures 1 and 2; Cannat et al., 1995). Together, these two ODP expeditions demonstrated that the internal stratigraphy of the lower oce anic crust at slowspreading ridges is gov erned as much by the dynamic processes of alteration and tectonics as by igneous processes. More recently, ODP Leg 209 (Sites 1268-1275; Figures 1 and 2) returned to drill in the peridotiterich area around the 15°20'N fracture zone and revealed that the upper oceanic lith osphere in this slowspreading setting is primarily composed of peridotite and gabbro and that the seafloor is inundated with uncovered fault surfaces (Kelemen et al., 2004). This leads to the conclu sion that mantle denudation and plate spreading are accommodated by a com bination of highdisplacement, lowangle (socalled "rolling hinge") normal faults that lead to the formation of oceanic core complexes and secondary lower displacement normal faults (Schroeder et al., 2007) that in turn expose the observed ultramafic basement rocks.

IODP Expeditions 304, 305, and 357: Atlantic Massif Ocean Core Complex
IODP Expeditions 304, 305, and 357 spe cifically targeted those types of denuded fault surfaces and a related ocean core complex, the Atlantis Massif at 30°N, which is located at the inside corner of the intersection between the Mid Atlantic Ridge (MAR) and the Atlantis Fracture zone. Two holes were drilled during IODP Expeditions 304 and 305 at Site U1309 (Figures 1 and 2) into the foot wall of the detachment fault (Blackman et al., 2006(Blackman et al., , 2011. This work was contin ued during IODP Expedition 357, which drilled a series of shallow holes into the Lost City hydrothermal field using sea bed rockdrills (FrühGreen et al., 2016). Based on the common occurrence of ser pentinized mantle peridotite along the south flank of the southern ridge as well as geophysical studies (e.g., Blackman et al., 1998Blackman et al., , 2002, fresh mantle peridotite was predicted to occur at reasonably shal low depths (~800 mbsf), allowing drilling to access samples of the mantle for the first time (Canales et al., 2004;Blackman et al., 2011). In stark contrast to geophys ical predictions, Hole U1309D sampled a 1,415 m long section of gabbroic rocks in the Central Dome core of the Atlantis Massif, with 75% recovery, but no per idotitic lithologies were encountered ( Figure 5). Paleomagnetic data obtained from the IODP core samples indicated that the footwall of the detachment fault   et al., 2009;Blackman et al., 2011), consistent with the "rolling hinge" model (e.g., Wernicke and Axen, 1988;Buck, 1988). Only three thin (<1 m) intervals of ultramafic rocks, interpreted as resid ual mantle peridotites, were encoun tered, and they were intercalated within gabbroic rocks in the upper 225 m of the section (Tamura et al., 2008) in Holes U1309B and U1309D. If the small amount of serpentinized peridotite recovered from Hole U1309D is representative of the bulk makeup of Atlantis Massif, the potential of a bulk expansion during the serpentinization of such altered peridotite is not likely to contribute significantly to the uplift of the Central Dome (Blackman et al., 2011). It is interesting to note that the 16 holes drilled by ODP and IODP into the footwall have reached four dif ferent oceanic core complexes, including the Atlantis Massif, and gabbroic sections were encountered exclusively at all holes. These findings indicate that the domal morphology of the core complexes results from the exhumation and unroofing of large gabbroic plutons by the associated detachment faults (e.g., Ildefonse et al., 2007a). Future scientific ocean drilling into both in situ slowspreading oceanic crust and related oceanic core complexes is needed (1) to fully understand the rela tionship between tectonics and magma tism in oceanic crust formation, (2) to determine the importance of serpentini zation in the lower oceanic crust and upper mantle, and (3) to fully grasp how serpentinization affects the seismic char acter of the oceanic lithosphere and the nature of the Moho.

MOHO TO MANTLE-FUTURE AND ONGOING DRILLING EFFORTS
Despite the successes of drilling into oceanic crust formed at both fast and slowspreading centers, drilling through the Moho and into the upper mantle remains a longterm aspiration, dating to the first Project Mohole operations in 1961. The MoholetoMantle (M2M) proposal (Umino et al., 2012) rearticu lated the major planetary science goals that could be achieved by sampling in situ upper mantle peridotite and inves tigating the nature of the Mohorovičić seismic discontinuity using the riser drill ing vessel Chikyu. This ambition remains a flagship proposal for future Chikyu drilling and would require penetrat ing at least ~6,000 m of igneous oceanic crust formed at a fastspreading ridge and an additional ~500 m into the oce anic upper mantle.
To determine the best site for M2M drilling, a large number of factors must be considered (Ildefonse et al., 2010). Any appropriate site should be in the shallowest possible water depths, imply ing close proximity to the axis of an active fastspreading midocean ridge. On the other hand, the hole should also be in the coldest possible oceanic lithosphere, implying mature oceanic crust and thus located a significant distance away from an active fast spreading ridge. Balancing those two opposing constraints lim its potential M2M sites to three areas off the coasts of Hawaii, Baja California, and Costa Rica (Figure 6; Teagle and Ildefonse, 2011). All potential sites are in the Pacific because the oceanic crust there was cre ated faster than in other ocean basins. As described above, seismic and geologic studies indicate that fastspreading oce anic crust is relatively uniform and con forms most closely to the endmember Penrose model (Figure 3), making those sites ideal and possibly most represen tative of the general processes of ocean crust formation. Although a site survey was conducted off the coast of Hawaii in 2017 (Ohira et al., 2018) and funding for future site surveys on the Cocos Plate have been secured, realization of proj ect Mohole continues to require a major funding commitment and political and scientific will. While preparing for even tual M2M drilling, any other scientific ocean drilling expedition, specifically at sites where the Moho is apparently shal lower, may provide further insight into ocean crust architecture, the role of ser pentinization, and the significance of the seismic Layer 23 and Moho boundaries.
IODP Expedition 360 was the first leg of Phase I of SloMo (shorthand for "The Nature of the Lower Crust and Moho at Slower Spreading Ridges"), a multi phase drilling project that proposes to drill through the Moho at Atlantis Bank at the ultraslowspreading Southwest Indian Ridge (MacLeod et al., 2017). By penetrating this fundamental seismo logical boundary, SloMo is testing the hypothesis that the Moho, at this locality in particular and at slow and ultraslow spreading ridges in general, may repre sent an alteration boundary due to ser pentinization within the upper mantle, rather than an igneous crustmantle tran sition or a hard physical boundary. If the Moho represents the former and thus is a serpentinization front, the igneous crust/ mantle boundary could lie at any depth above the seismic boundary (MacLeod et al., 2017). IODP Hole U1473A (Figure 2) was drilled on the summit of Atlantis Bank during Expedition 360, 1-2 km away from two previous ODP holes: Hole 735B drilled during ODP Leg 118 in 1987 (Dick et al., 1999; and Hole 1105A drilled during ODP Leg 179 in 1998 . While exploring the lateral vari ability of the stratigraphy in comparison with Holes 735B and 1105A (Figure 5), the principal aim of Expedition 360 was to drill as deep as possible through lower crustal gabbro and leave a hole open and ready to be deepened during a second expedition. A target depth of 1,300 mbsf was estimated, derived from prior expe rience of drilling conditions at Atlantis Bank; however, Hole 1473A was only drilled to 789.7 mbsf and terminated in massive gabbro cut by isolated dikes (Figures 2 and 5; MacLeod et al., 2017). SloMo next will attempt to reoccupy and deepen the hole with the overall goal of penetrating the crustmantle tran sition, which is believed to be as much as ~2.5 km above the Moho; additional drilling, potentially using the riser vessel Chikyu, is likely to be necessary to pene trate the Moho itself, at ~5 km below the seafloor (MacLeod et al., 2017).
Another approach to sampling upper mantle materials is to drill and core fresh lower igneous crust and the underlying uppermost mantle peridotite, as accreted during the initiation of a subduction zone. A prime IODP focus has been the study of subduction initiated around ~52-48 mil lion years ago at the IzuBoninMariana trench (e.g., Ishizuoka et al., 2011;Reagan et al., 2017Reagan et al., , 2019; see also Arculus et al., 2019, in this issue), where gabbroic and ultramafic rocks are exposed on the land ward slope of the Bonin Trench in the Northwest Pacific ( Figure 6). Drilling at this location provides future opportuni ties to realize a key objective of the M2M mantle drilling (Michibayashi et al., 2016) that differs fundamentally from the M2M itself and SloMo, which both focus on the formation of the oceanic crust during seafloor spreading.

CONCLUDING REMARKS
For 50 years, scientific ocean drilling has contributed significantly to our under standing of the variability in the archi tecture of oceanic lithosphere. The style of accretion critically depends on the bal ance between magma production, hydro thermal cooling, and tectonics, which to a first order is related to spreading rate. Seismic, bathymetric, and marine geo logical observations indicate that oceanic crust formed at fast spreading rates (with full rates >80 mm yr -1 ) has a relatively constant architecture, compared to crust formed at slow to ultraslow spreading rates (<40 mm yr -1 ), and is similar to the Penrose model for ophiolites (Ildefonse et al., 2014). Scientific ocean drilling at ultraslow spreading centers and at oce anic core complexes has shown that their crustal architectures are very different and that serpentinization likely plays a promi nent role in changing the nature of those crustal sections. Deeper drilling efforts to penetrate the coremantle boundary and the Moho remain the missing piece of the puzzle to help us advance our under standing of ocean crust formation and mantle dynamics. For the upcoming next generation ocean drilling scientists, we end with the following quote by Bahcall (1990): "I believe that the most import ant discoveries will provide answers to questions that we do not yet know how to ask and will concern objects that we can not yet imagine. "