RUSSIAN JOURNAL OF EARTH SCIENCES VOL. 7, ES4001, doi:10.2205/2005ES000181, 2005

Discussion

The general characteristics of rocks of the silicic Fe-Ti-oxide association.
The petrography and geochemistry of the rocks indicate that they probably comprise two groups of plutonic rocks: (i) magnesian (primitive) troctolites, gabbro, and olivine gabbro, whose geochemistry is close to that of MORB; and (ii) variable kaersutite-bearing Fe-Ti-oxide rocks, including diorites and trondhjemites, which show some characteristics close to those of E-MORB but notably differ from them in having a deficit in LREE, Zr, Th, Hf, K, and, sometimes, Ba. A separate group is composed of the basalts, which generally have trace-element compositions close to those of MORB but have positive Pb anomalies and, in four samples, also positive Cs and Rb anomalies. The U-Pb zircon of a sample of one of these rocks (cataclased hornblende gabbronorite, sample I1028/1) is 97.42 pm 0.13 Ma (see above). [Sharkov et al., 2004a].

[57]  It was mentioned above that most rocks of the Fe-Ti-oxide association in slow-spreading mid-oceanic ridges are thought to have been produced by the fractional crystallization of MORB-type melts [Dick et al., 1992]. However, these rocks notably differ from the final products of the crystallization differentiation of tholeiitic basalts, which are thoroughly studied in continental layered intrusions. The ore mineral of these rocks is titanomagnetite, and they contain practically no magmatic hornblende but bear orthoclase and are typically enriched in trace elements, particularly LREE [Wager and Brown, 1968]. The interpretation of the Fe-Ti-oxide rocks as late derivatives of MORB is also at variance with their wide occurrence in the Sierra Leone MAR segment in the form of plutonic, subvolcanic, and volcanic rocks.

[58]  Furthermore, the petrology and geochemistry of the Fe-Ti-oxide rocks cannot be described within the framework of any currently existing systematics. Although these rocks are often anomalously rich in Ti, Nb, and Ta, as is typical of within-plate melts related to the activity of deep mantle plumes, they are poor in incompatible elements, including LREE. These rocks are characterized by the presence of orthopyroxene, pigeonite-augite, and magmatic hornblende, i.e., minerals typical of the calc-alkaline series, which were produced by melts saturated with silica and water. Elevated water contents in the melts also follow from the separate crystallization of their Fe-Ti oxides (see above). In other words, the petrography and geochemistry of these rocks indicate that they compose a separate magmatic association, which differs from both within-plate rocks and those of the calc-alkaline series but displays certain features of both.

[59]  The possibility of the occurrence of such magmas in oceanic environments follows from data obtained by studying melt inclusions in chromite from the dunites of the transitional zone between mantle harzburgites and crustal cumulates in the Oman ophiolitic association [Schiano et al., 1997]. The inclusions have a composition of magnesian (mg# = 63.5-66.8) hypersthene-normative basalt and show depleted REE patterns with significant Nb anomalies, as is typical of melts of suprasubduction provenance, but, in contrast to the latter, high TiO 2 concentrations (of about 3 wt %).

[60]  Closely related to the Fe-Ti-oxide association are ultramafic cumulates (harzburgites and lherzolites). As was demonstrated above, these ultramafic rocks display geochemical characteristics notably different form those of mantle rocks, and their mineralogy suggest the presence of at least two independent mineral assemblages. The earlier of them is includes Fe- and Ti-richer spinel, which makes it notably different from mantle peridotites through which basaltic melts percolated (the latter rocks show the opposite relations [Piccardo et al., 2004]). Moreover, the rocks dredged at sites within the Sierra Leone MAR segment (sites I1022, I1025, I1032, I1060, I1063, I0168, L1122, and others) include, along with these rocks, also norites, gabbronorites, olivine norites, Opx -bearing troctolites, and bronzitites. All of these features suggest that the area includes a series of intrusive rocks (cumulates) all of which contain low-Ca pyroxene and, often, also magmatic amphibole and Fe-Ti oxides. We propose that this group of intrusive rocks, whose composition varies from ultramafic to acid, should be recognized as an individual silicic Fe-Ti-oxide series. Its subvolcanic analogues seem to be the ilmenite-hornblende dolerites, and volcanic counterparts are the aphyric Fe-Ti-oxide and plagioclase- and olivine-plagioclase-phyric basalts with abundant disseminated ilmenite, which were dredged at these sites.

[61]  The positive Eu and Sr anomalies in the gabbroids of the Fe-Ti-oxide series and negative anomalies in its intermediate-acid members suggest that these elements were fractionated during plagioclase crystallization. Analogous but weaker anomalies of opposite character are also typical of U and Pb, whereas Th, conversely, is enriched in the intermediate-acid rocks. This suggests that the Fe-Ti-oxide gabbro, diorites, and trondhjemites are indeed derivatives of melts other than MORB. As was mentioned above, the rocks of this series and their supposed derivatives plot in classification diagrams within a separate field and but not within the fields of any other known series.

[62]  For the equatorial MAR segment, the occurrence of glasses whose Nd-Sr-Pb systematics testify that they were produced by the mixing of isotopically distinct mantle sources: HIMU (i.e., that containing excess radiogenic Pb, supposedly related to a subduction component [Zindler and Hart, 1986]) and the depleted mantle: was first suggested by Schilling et al. [1994]. These researchers explained this phenomenon by the effect of the hypothetical Sierra Leone plume, which should have partly assimilated the buried subducted material. A complex composition of the melt sources beneath this MAR segment also follows from geochemical data, which point to the existence of three or four mantle reservoirs [Hannigan et al., 2001]. This is in general agreement with our results, according to which the area includes at least two geochemical types of ancient plutonic rocks, and a third one is represented by fresh basalts that are close to E-MORB but show some characteristics distinct from them. At the same time, no magmatic rocks ever found in the area can be related to the activity of plumes, neither are there any traces of subduction, and the characteristics of the silicic Fe-Ti-oxide association are at variance with its affiliation with any of the currently known magmatic series.

[63]  The conclusion that such rocks have an unusual genesis was arrived at by Benoit et al. [1999] based on studying gabbronorite cumulates in the Oman ophiolites, and by Holm [2002], who examined the Sr-Nd-Pb systematics of Fe-Ti-oxide gabbro and diorites that occur as dikes in the Southwest Indian Ridge (Atlantis II Deep, Hole 735 B, ODP Leg 176). According to these geologists, melts of this type have a complicated genesis, including the mixing of MORB-type magmas and LREE-depleted melts.

The magmas of the silicic Fe-Ti-oxide association as hydrous melts.
The facts that these rocks ubiquitously contain magmatic hornblende and are characterized by the separate crystallization of magnetite and ilmenite serve as direct evidence that the melts contained water. Judging from the Nd-Sr isotopic data, this water was of marine provenance [Sharkov et al., 2004a]. This puts forth the question as to how seawater could be brought into the magma-generating region or the already-solid intrusive chamber where crystallization differentiation took place. Water could hardly percolate into it, because heated water or a vapor-water mixture should have ascended along fractures in the host rocks (as is commonly observed in modern volcanic areas) and give rise to various fumaroles in continental environments and to a diversity of black smokers in the ocean. Relations between magmatic melts and water can be vividly illustrated by the example of pillow lavas that were erupted on the seafloor: their chill glasses contain practically no seawater [Simonov et al., 1999], and the basalts themselves were altered only locally along contraction cracks or in the course of halmyrolysis.

[64]  We believe that seawater was brought into the magmatic melt mostly during the process of magma generation, due to the remelting of hydrated mafic-ultramafic rocks in the upper oceanic lithosphere. As was mentioned above, these ultramafic rocks are commonly serpentinized and the gabbroids are variably altered with the development of fibrous actinolite, veins of chlorite, prehnite, epidote, and numerous hydrothermal-metasomatic zones of variable thickness. The LOI values (mostly, the water losses) in the chemical analyses of the gabbroids vary from 1 wt % to 2.5 wt % (Table 8), whereas the analogous values for their serpentinized analogues are as high as 9-17 wt % at average values of 13-14 wt % [Simonov et al., 1999]. When a hot mantle diapir (lithospheric plume) intrudes such hydrated lithospheric material, it is reasonable to expect that this material starts melting, because the presence of water in the rocks notably decreases their solidus temperatures. These magmas of new generation should obviously have some specific geochemical characteristics, which distinguish them from melts that came directly from the mantle, because the composition of these newly formed melts was notably affected by the fractionation of elements during melting and crystallization in the presence of an aqueous fluid.

Geochemical aspects of the melting of the hydrated oceanic lithosphere.
These aspects are now actively explored in the context of the genesis of the calc-alkaline series via the melting of subducted slabs that consist mostly of MORB-type protoliths. As follows form experimental data, the partial melting of these water-saturated basalts is possible at temperatures higher than 750o C [Peacock et al., 1994]. According to the experimental data in ([Tatsumi et al., 1986], these partial melts should have had high concentrations of Ti, Nb, and Ta. The data presented in [Schiano et al., 1995] suggest that these melts should also have been enriched in silica. This in good agreement with our petrographic and geochemical data on the rocks of the silicic Fe-Ti-oxide series. Although these rocks are notably different from suprasubduction rocks of the arc calc-alkaline series, they display certain similarities with them, which are caused by the role of water in the genesis of the parental melts.

[65]  Important information on the behavior of trace elements during the dehydration of the oceanic crust in subduction zones was obtained by Becker et al. [2000] for temperatures of 600-900o C, at which hydrated basites are melted [Peacock et al., 1994]. These researchers have demonstrated that the elements most significantly removed from the protolith are Pb and U, which are followed by Nb, K, Rb, Ba, and, then, by Cs and Sr. An analogous situation exists with serpentinites. The experimental data on trace elements in fluids released from these rocks during deserpentinization indicate that these fluids are enriched in Cs, Be, Li, Cl, As, Sr, Rb, Ba, and Pb, which were borrowed by serpentinites from seawater when the peridotites were altered at the seafloor [Tenthorey and Hermann, 2004].

[66]  All of these elements mobile during melting in fluids should have obviously dissolved in the melt and caused the respective positive anomalies in it relative to MORB. The newly generated melts were not significantly enriched in K likely because of the very low concentrations of this element in the source material (serpentinites, primitive gabbroids, and troctolite; see above). In the course of hydration, Nd, Sm, Zr, and Y are practically immobile, and, thus, their concentrations in the melt did not notably change compared with those in the protolith. Most REE were also not fractionated in the coarse of this process, because the character of their distribution in rocks remains the same as in MORB, although their total contents notably increased, particularly in the diorites and trondhjemites. The deficit in LREE seen in the chondrite-normalized patterns of the latter two rocks is notably lower (Figure 17), which suggests that these rocks were generated by crystallization differentiation.

[67]  In this context, it is worth mentioning that practically all of our samples of the basalts show positive Pb anomalies, and basalts from sites I1026 and I1027 additionally have Cs and Rb anomalies. Conceivably, these basalts underwent mixing with partial melts derived form the hydrated mafic rocks or are themselves basalts of the second generation.

[68]  This mechanism of magma derivation related to the remelting of mafic and ultramafic rocks of the hydrated oceanic lithosphere and the associated extraction of more ancient Pb from these rocks provides a plausible clue to understanding the HIMU isotopic source, which was identified MAR. In this situation, there is obviously no need to conjure some hypothetical subduction material to interpret data on the melt sources in these specific active magmatic structures.

[69]  The occurrence of cumulus ultramafic rocks among the rocks of the Fe-Ti-oxide association suggests that melting affected not only the altered mafic rocks of the oceanic crust but also serpentinites that developed after ultramafics in the upper oceanic lithospheric mantle. Experimental data indicate that the melting of these rocks at 1-3 kbar (i.e., at depths of 3-10.5 km) can begin already at temperatures of about 930-980o C [Kushiro et al., 1968]. At pressures of about 10-12 kbar and 12 wt % H2O, the melting temperatures of serpentinites increase to approximately 1115oC. Because of the increase in the Si/(Mg + Fe) ratio, the newly formed melts are quartz-normative [Gaetani and Grove, 1998]. This implies that the solidus temperatures of the serpentinites were significantly lower than the asthenospheric mantle temperatures throughout the whose range of depths possible under these conditions.

[70]  Inasmuch as the newly formed melts were in equilibrium with normative olivine and pyroxenes, their first cumulates should have been of ultramafic composition. As follows from experimental data on the Fo- Di-SiO2 system [Kushiro, 1969], which describes the crystallization of such melts, their evolutionary trend should shift (due to the presence of water) toward SiO2 and result in a granitic liquid as the final derivative.

[71]  Available petrological and geochemical data provide evidence that the melts of the silicic Fe-Ti-oxide series could be generated by the remelting of the hydrated oceanic lithosphere, since they are oceanic analogues of mantle-crustal melts. Because of this, these melts share some characteristics with the calc-alkaline series but differ from it in having strongly subordinate amounts of rocks of intermediate and acid composition, which are predominant in arc series. These differences are obviously caused by the fact that the subducted slabs contain much sediments and continental crustal material from the backarc space [Sharkov, 2003b; Wilson, 1989], which are absent from mid-oceanic ridges. The geochemical differences are no less important: our rocks are generally poor in incompatible elements and have elevated to high concentrations of Fe, Ti, Ta, and Nb, which are atypical of calc-alkaline magmas. Obviously, the latter feature is related to the derivation of the magmas from the rocks of the subducted slab under high pressures, when the residue contains rutile, the main concentrator of Ti, Ta, and Nb [Peacock et al., 1994].

Possible mechanism forming secondary magmatic systems in the hydrated oceanic lithosphere.
What could cause the processes of melrting in this lithosphere? Obviously, this could be only the emplacement of new plumes. Geochemical data indicate that these could not be deep plumes like those maintaining the development of oceanic islands. Available data indicate that the composition of these plumes was most probably close to the MORB source, i.e., they were protuberances on the surface of large asthenospheric lenses beneath mid-oceanic ridges. According to geophysical evidence, such lenses have thicknesses of about 300 km, which makes it possible to maintain the activity of these global structures [Fowler, 2004].

[72]  In papers dealing with the possibility of melting of the hydrated oceanic lithosphere during the emplacement of mantle diapirs, it is usually assumed that this melting occurs at contacts with these diapirs [Benoit et al., 1999; Ross and Elthon, 1996]. This is, however, hardly possible because (i) the marginal parts of diapirs (plumes) are cooled in contact with the colder host rocks, and (ii) the mechanism of conductive heat transfer is extremely inefficient and cannot maintain the stable development of magmatic systems. In this sense, a much more promising mechanism is melting above large magma chambers that develop immediately at the plume roof and make it possible underplating, a process responsible for the formation of the lower mafic crust of continents [Rudnick, 1990].

[73]  The most probable mechanism that can enable this process is zone melting, which was proved efficient for the formation of the melts of the Paleoproterozoic silicic high-Mg series [Sharkov et al., 1997]. In essence, this process is the ascent of a chamber of a high-temperature mantle magma through the mafic lower crust by means of melting its roof rocks and the simultaneous crystallization of the melt near the chamber bottom because of the difference between the adiabatic gradients and melting points [Sharkov, 1980]. Since the temperature of the parental mantle magmas of the MORB type is 1260-1280o C [Sobolev et al., 1988] or are even as high as 1300-1400o C [Ryabchikov, 2003], i.e., much higher (by at least 300-400o C) than the solidus temperature of serpentinites and hydrated mafic rocks (see above), this heat is sufficient to induce the melting of the roof rocks.

2005ES000181-fig22
Figure 22
[74]  The ascent of such a new plume should have stopped when it reached the limit of its buoyancy in the low-density serpentinized upper mantle. After this, the plume heads started spreading laterally. This process was associated with adiabatic decompression and the melting of the ultramafic plume material. The newly formed melt first accumulated underneath the cooled roof of the plume and then migrated upward along arrays of extension fractures and formed large mantle chambers. Data on ophiolitic associations indicate that such a chamber originally developed between the plume head and the overlying older lithosphere [Sharkov et al., 2001]. Due to convection, heat was lost from this chamber predominantly through its roof, which, in turn, should have inevitably brought about its melting because of the relatively low solidus temperatures of the hydrated roof rocks. The relatively cold and partly molten material of the roof should have sunk to the inner parts of the chamber or has completely or partly dissolved in the melt (Figure 22). This dissolutions seems to have not always been complete, as follows from the occurrence of foreign assemblages among the minerals (see above). The structural water of the hydroxyl-bearing minerals in the melting rocks should have been actively consumed by the melt and maintain its elevated water contents.

[75]  In contrast to solidification processes, which proceed from cold contacts toward the central portion of cooling magmatic chambers (intrusions) that are isolated from the host rocks by marginal zones [Sharkov, 1980; Wager and Brown, 1968], the processes of melting should have proceeded in the opposite direction (from the center outward). Because of this, the melt received components not only from the melting roof rocks of the chamber but also from rocks in its periphery, as well as fluid components from the strongly heated rocks in the remote periphery. This likely predetermined the unusual features of these melts, their enrichment in components most mobile under these conditions, such as silica, Ti, Pb, and Cu.

2005ES000181-fig23
Figure 23
[76]  According to the mechanism of zone melting, the melting of the roof should be associated with the crystallization of the most refractory minerals near the chamber bottom. This mechanism maintains the continuous ascent of the chamber through the lithosphere and the systematic involvement of progressively higher lithospheric levels (including layer 3 of the oceanic crust) in the melting process. The long-lasting character of this process should have been maintained by the periodic arrival of new portions of fresh melt from the mantle magma-generating zone, as was established, for example, for the Paleoproterozoic [Sharkov, 2003a]. The surmised inner structure of such a magmatic system is illustrated in Figure 23.

[77]  The ascent of the magma chamber and the associated large-scale assimilation of the material of the overlying lithosphere resulted in the continuous enrichment of the melt in the melting products of this lithosphere and in the products of crystallization differentiation. Inasmuch as the melt composition in the chamber was constantly equalized by convection (which homogenized the whole melt volume), these changes should also have affected the composition of the crystalline phases settling near the chamber bottom. Consequently, the ascending chamber left behind a trail in the form of a series of specific cumulates, ranging from ultramafic to mafic rocks, with the inner structure of the series resembling that of layered intrusions. Indeed, as was mentioned above, MAK acoustic sounding data indicate that the flanks of the Markov depression look like a cross section through a large layered intrusion. Low-melting components should have enriched the stirred layer of the melt and produced Fe-Ti-oxide rocks and trondhjemites late in the evolutionary course of such a magmatic system.

[78]  Unlike typical layered intrusions, this mechanism of the origin of the layered series implies that the cumulate and the crystallization products of the melt could include the disintegrated incompletely melted material of the ancient continental lithosphere, which consisted of various mafic and ultramafic rocks. The U-Pb age of the latter could be as old as hundreds of million years [Pilot et al., 1998; Sharkov et al., 2004b]. This suggests that the third layer of the MAR oceanic crust was formed for a long time and is heterogeneous due to the multiple emplacements of asthenospheric plumes into the oceanic lithosphere. When this lithosphere was not hydrated, it gave rise to melts of the MORB type, and, conversely, the silicic Fe-Ti-oxide series was derived from the lithosphere affected by low-temperature alterations.

[79]  A portion of melt from a large chamber cold also separate from it and produce small individual intrusions of gabbroids among serpentinites, as was determined in MAR at 15o N [Cannat and Casey, 1995; Cannat et al., 1997]. Judging from the descriptions of the rocks (gabbroids with brown hornblende, ilmenite, apatite, and zircon; trondhjemites), these gabbroids should affiliate to the series considered in this publication. Analogous rocks (Fe-Ti-oxide hornblende gabbro and gabbronorite; our unpublished data) were dredged from the axial zone of MAR during cruise 22 of the R/V Professor Logachev, during which the Ashadze hydrothermal field was discovered at 13o N, 44o52 prime W. The field contains massive sulfide mineralization among serpentinized peridotites [Bel'tenev et al., 2004]. A fairly similar situation seems to exist at other MAR hydrothermal fields with ore mineralization [Bogdanov et al., 1997].

Silicic Fe-Ti-oxide series and ore-forming processes.
The solubility of fluid components, particularly water, in magmatic melts is known to decrease with decreasing pressure and, correspondingly, the solidus temperatures of the melts increase. Consequently, the ascent of hydrous melts is associated with water release from them, with this water becoming excessive under the new conditions. Upon reaching the hydrous solidus, the melts start to rapidly crystallize, and, hence, magmas of this type can only rarely reach the surface [Popov and Bogatikov, 2001]. According to experimental data, a particularly drastic decrease in the water solubility in magmatic melts occurs at pressures of approximately 1 kbar, i.e., at depths of about 3-4 km [Kadik, 1991]. This seems to be the upper limit for the ascent of most magmas of the Fe-Ti-oxide series, whose crystallization products usually occur as intrusions but are very rare in the form of lavas.

[80]  The data and considerations presented above may also shed light on the genesis of mineralized fluids in the oceanic crust. The origin and development of hydrothermal systems is now thought to be related either to the circulation of waters near the surface when these waters are heated near magma chambers [Grichuk, 2000] or to the serpentinization of peridotites [Bogdanov et al., 2002]. None of these models can, however, explain the occurrence of significant amounts of Pb, Zn, and Cu in these hydrothermal waters. Nevertheless, exactly these components should be contained in predominantly aqueous solution that escaped from the water-bearing melts of the silicic Fe-Ti-oxide series during decompression, and, conceivably, they also played a decisive role in the origin of ore-forming hydrothermal systems.

[81]  Indeed, most sulfide occurrences in the MAR axial zone, including its black smokers, are spatially restricted to areas with rocks of the Fe-Ti-oxide association. This pertains to both the massive sulfide ores of hydrothermal-metasomatic genesis among altered gabbroids in the Markov depression [Pushcharovskii et al., 2002] and the hydrothermal fields with massive sulfide mineralization in serpentinized peridotites, such as the Ashadze field in the axial MAR zone. A similar situation also seems to occur at many other mineralized hydrothermal fields in MAR [Bogdanov et al., 1997].

[82]  In conclusion, it is pertinent to mention that the recognition of a principally new magmatic association among the rocks of the slow-spreading MAR can be very important for understanding deep petrogenesis in similar ridges, provides insight into processes forming ore sulfide mineralization in them, and facilitates searches for black smokers and related mineral deposits on the seafloor.


RJES

Citation: Sharkov, E. V., N. S. Bortnikov, T. F. Zinger, and A. V. Chistyakov (2005), Silicic Fe-Ti-oxide series of slow-spreading ridges: petrology, geochemistry, and genesis with reference to the Sierra Leone segment of the Mid-Atlantic Ridge axial zone at 6° N, Russ. J. Earth Sci., 7, ES4001, doi:10.2205/2005ES000181.

Copyright 2005 by the Russian Journal of Earth Sciences

Powered by TeXWeb (Win32, v.2.0).