Mantle-Derived Magmas II

Transcription

1 Mantle-Derived Magmas II 1. Enrichment in Ocean Island Basalts and E-MORB. We saw in the last lecture how OIB and E-MORB have enriched incompatible element systematics compared to depleted N- MORBs. How might these enrichments arise? 1.1. Partial Melting of Depleted/Fertile Mantle Incompatible elements are concentrated into a magma on melting and are those elements that are not easily incorporated into the mineral phases of the protolith rock (i.e. peridotite). At small degrees of partial melting the most incompatible elements are concentrated in the magma, however, with increasing melting less incompatible elements are forced to partition into the melt. The magma, therefore, becomes less incompatibleenriched with increasing melting (increasing temperature above the melting point) because less incompatible elements are being added to dilute the more compatible elements. Magmas generated by small degrees of partial melting are more enriched in incompatibles than magmas generated by larger degrees of partial melting. Although variations in partial melting change the enrichment of incompatible elements in the generated magma they cannot explain the enrichment of OIB and E-MORB. These are particularly enriched in the LREE rather than the most incompatible elements Isotopic Evidence Variations in isotopic ratios are often used to study the origins of magmas. Two isotopes of the same element, particularly the heavier radiogenic nuclides, behave in exactly the same way during igneous processes such as melting and crystallisation, because they are the same size, charge and their masses are only slightly different. The ratio of two isotopes of the same element will, therefore, initially be the same in a magma as in its source rock. Changes in isotope ratios occur due to (a) radioactive decay and (b) the mixing of materials (e.g. magmas) from sources (i.e. reservoirs) which have different ratios. Isotopic ratios thus provide information on the sources of magmas and fluids. 1.2.a. Sr isotopes Strontium isotope ratios are particularly important in the study of igneous rocks. Rb decays with a half-life of 4.88x10 10 yr into Sr. 86 Sr is a stable isotope and is not produced by radioactive decay. Over the lifetime of the Earth the Sr/ 86 Sr ratio of the bulk Earth has, therefore, increased (see graph opposite showing Sr/ 86 Sr against time). The rate of the increase with time depends on the starting Rb/Sr (because this determines the amount of radioactive Rb available to make Sr through radioactive decay). If the starting Rb/Sr was higher the rate of increase of Sr/ 86 Sr would also be higher (i.e. a steeper slope). Although igneous processes (melting and crystallisation do not separate Sr and 86 Sr, and thus do not change the Sr/ 86 Sr ratio, they do change the Rb/Sr ratio because Rb is more incompatible than Sr. A magma will, therefore, contain a higher Rb/Sr ratio than it s source rock since melting concentrates the most incompatible elements (e.g. Rb) relative to less incompatible elements (e.g. Sr). Because the magma has removed more Rb from the source rock than Sr during partial melting, the residual rock, left over after extraction of the magma, will have a lower Rb/Sr ratio than the original rock before melting. The change in Rb/Sr ratios means that Sr/ 86 Sr changes along different slopes (see above). N-MORB has Sr/ 86 Sr ratios of to lower than bulk Earth (fertile mantle) value of indicating it is derived from lithospheric mantle depleted by the extraction of basaltic partial melts (as indicated by the trace elements). The Rb/Sr ratio of MORB of 0.01 suggests that depletion occured 1.5 Ga. However, this is over-simplified since

2 we know that the lithosphere is being depleted continuously by the production of MORB and depletion may, therefore, have begun before. This possibly relates to the initiation of modern plate tectonics ~2.0 Ga. Prior to this the geothermal flux was much higher due to decay of radiogenic isotopes. Geothermal flux was, for example, twice as large as today 2.5 Ga. OIB (and to a lesser extent E-MORB) have higher Sr/ 86 Sr ratios than N-MORB, however, some have lower ratios than bulk Earth (fertile mantle). The Sr isotopes of OIB and E-MORB indicate the enrichment of the depleted lithosphere by incompatibles (i.e. Rb). Another common way of displaying isotope data is to plot the ratio of the daughter nuclei/stable nuclide against the ratio of the parent element/daughter element (i.e. Sr/ 86 Sr versus Rb/Sr). These are known as isochron plots (see above). The bulk Earth had an original Sr/ 86 Sr ratio of around Ga. Imagine that at its formation 4.5 Ga that the Rb/Sr ratio was different throughout the planet. At that time all these rocks would plot along the isochron labelled as at 4.5 Ga. With time, however, Rb will decay into 86 Sr and the ratio of Sr/ 86 Sr will increase. The more Rb the rock contained the larger the increase in the ratio. After 1.5 bn yr all the rocks would plot along the isochron labelled Ga. With more time more Rb will decay producing even more Sr. Today, 4.5 bn yrs later, all the rocks would plot along an isochron called the Geochron if they had not been disturbed (i.e. had Rb removed from Sr by melting). Any rock plotting on the low Rb/Sr side of the geochron such as MORBs must have originated from a source that has been depleted in Rb (e.g. through the extraction of basaltic magma). Any rock plotting on the high Rb/Sr side of the geochron originated from a source that has been enriched in Rb. 1.2.b Neodymium Samarium Isotopes Another system useful for examining the sources of mantle-derived magmas is Nd-Sm. Both these elements are REE and 147 Sm decays to 143 Nd with a half life of 1.06x10 11 yr. The stable isotope is 144 Nd. Neodynium-Sm isotopes are used in exactly the same way as Rb-Sr, however, Samarium the more incompatible than Nd (the opposite way round to Rb-Sr). Magmas (and rocks) derived from depleted mantle will have higher 143 Nd/ 144 Nd than rocks derived from a fertile undepleted mantle. MORB has 143 Nd/ 144 Nd ratios of to higher than bulk Earth at Nd-Sm and Rb-Sr isotopes can be used together to provide information on the source materials for mantle enrichment. The Nd- Sm and Rb-Sr systematics of MORB, OIB, Continental Crust and Oceanic Sediments lie on a mixing line between depleted mantle and oceanic sediments. The isotopic systematics of OIB could, therefore, be explained by adding components derived from oceanic sediments to the depleted mantle. Ocean islands are, however, usually not located close to subduction zones. One recent theory on how the enriched OIB mantle source could be generated suggests that oceanic sediment (derived from the continents) is carried deep into the mantle by the subduction process. Because subducted lithosphere is cold and dense it may continue to sink into the mantle and eventually reaches the core/mantle boundary. In doing so, the subducted material carries some oceanic sediment that is enriched in incompatible trace elements and has high ratios of Sr/ 86 Sr and low ratios of 143 Nd/ 144 Nd. This enriched component mixes with the mantle near the core/mantle boundary to produce an enriched mantle. The enriched mantle eventually heats up as a result of heat released from the Earth's core. When it is sufficiently hot it begins to rise in narrow plumes that channel the enriched mantle upward and produce hotspots as they rise through the asthenosphere. Melting in the rising plume generates magmas that intrude and enrich the depleted lithospheric mantle. 2. Diversification of Basaltic Magmas We ve already seen that in Hawaii and Iceland that tholeiitic basalt, alkali basalt and basanite magmas can evolve to produce a wide variety of rocks through repeated extrusion of lava from a magma chamber in which crystal fractionation is occurring. The process of crystal settling leads to the accumulation of crystals at the base of the magma chamber and is preserved within layered intrusions. These provide natural experiments by which we can examine crystal fractionation in detail Crystal Fractionation and Layered Intrusions. Layered basic intrusions occur throughout the geologic record on the continents. They are generally large, funnel-shaped bodies of crystalline igneous rock that have an overall composition close to that of a mid-ocean ridge basalt. More

3 importantly, most of them crystallized in situ from a single batch of very primitive magma. These intrusions illustrate the behaviour of an ideal magma chamber in which a single batch of magma is allowed to cool and fractionate. 2.1.a Skaergaard Intrusion The Skaergaard Intrusion in East Greenland is a classic example of a layered basic (mainly gabbroic) intrusion and forms a skewed lopolith, 500km 3 in volume, that was a subvolcanic magma chamber. A reconstruction of the intrusion (before folding and erosion) is shown opposite. It shows that the margins of the intrusion are surrounded by a border series (Upper and marginal) which in the marginal series is finergrained massive gabbro formed as a chilled margin around the magma chamber. The upper border group is coarser-grained and massive, however, a finer-grained early upper border group was probably formed above this as a chilled margin. Such chilled margins are seen on most layered intrusions and differentiated sills. The central region contains a layered series that is split into Upper, Middle and Lower Zones on the basis of its petrology and consists of a series of gabbros. The layered series formed by the accumulation of crystals settling out of the magma chamber (i.e. they are cumulates). Three types of layering are found: (1) cryptic layering, in which the compositions of cumulus phases that are members of a solid-solution series, such as olivine, pyroxene and plagioclase change through the intrusion, (2) phase layering, where certain minerals appear at particular levels in the intrusion (i.e. olivine is absent from the Middle Zone, and (3) rhythmic layering, repeated units mm to m in thickness of normal graded gabbro with mafics concentrated at the base. 2.1.b Crystal Fractionation of Skaergaard The change in mineral composition with depth into the intrusion relates to changes in the composition of the magma due to crystal fractionation (crystallisation and settling). The layered series shows an increase in Fe/Mg ratio of olivine and pyroxene and an increase in Na/Ca ratio in plagioclase (shown above). This change occurs because at high temperatures olivine and pyroxene that crystallise from a magma are Mgrich and plagioclase is Ca-rich. When these crystallise they remove more Mg and Ca from the magma than Fe and Na and so the magma becomes more Fe- and Na-rich. With cooling (as the magma chamber becomes smaller due to the accumulation of crystals) more Fe-rich and Na-rich olivine/pyroxene and plagioclase crystallise. Once the magma has become very rich in Fe then Fe-rich phases (such as fayalite and iron-oxides) begin crystallising. These phases are silicapoor and their crystallisation makes the magma become more silica-rich. In Skaergaard the UZ layered series is dominated by ferrodiorite, a fayalite-bearing intermediate rock, rather than gabbro. Small bodies of acid granophyre (pegmatites) are also present and illustrate that crystal fractionation of basaltic magmas can produce intermediate and acid magmas after more than 95% of the melt has crystallised. The overall trend of the magma composition is best seen on the ternary diagram above and is similar to Hawaiian and Icelandic magma series. In many basic magmas (although perhaps not in Skaergaard), olivine is crystallises at the highest temperature, and is then joined by pyroxene and finally by plagioclase. As a result the basal cumulates are dunite (formed by settling of olivine) with peridotite (olivine + px) above and finally gabbro as plagioclase begins to crystallise and settle. This sequence is seen in other layered intrusions and at the base of ophiolite sequences (formed by crystallisation of a magma chamber beneath the Mid Ocean Ridges). 2.1.c. Rhythmic & Phase Layering Phase layering in Skaergaard occurs when the change in the composition of crystallising phases in the magma makes these phases unstable. Olivine disappears in MZ because Mg-rich olivine reacts with the magma to produce orthopyroxene (see system forsterite-silica lecture 6). Fayalitic (Fe-rich) olivine, however, doesn t react and thus olivine reappears once the magma has become richer in Fe.

4 The rhythmic layering originates due to convection currents, density and size controlling settling rates and pulses of crystallisation. Crystals settle when their density is higher than the surrounding magma, however, the speed that they fall also depends on their size, with smaller crystals settling the slowest (like in normal grading). Settling of crystals of different sizes and shapes could explain the observed layering if crystallisation did not occur continuously but in pulses. A pulse of crystallisation could occur, for example, due to supercooling and sudden nucleation. Once the crystals start growing they remove the nuclei from the magma which then have to build up again producing a pause in crystallisation. The formed batch of crystals can then settle to form a normally graded layer. Plag crystals are less dense than basaltic liquid and should float, however, they may settle if they collect together with denser minerals in rafts. Convection also affects crystallisation and crystal settling since it produces rising and falling currents. With velocities of several m per day rising currents can prevent crystals from settling. Crystals in the upper border series of Skaergaard, which grew downwards, must have been carried there (upwards) by convection currents. The upper border series shows increases in Fe-content downwards since it grew by attachment of crystals from the top down. Crystallisation is also more likely to occur in the falling currents because with depth the hydrostatic pressure increases and the liquidus temperature rises (causing further crystallisation). Slight reverse zoning in some plag crystals in the LZ of Skaergaard may be caused by changes in pressure experienced by crystals caught in convection currents. Pulses of crystallisation in falling convection currents probably explain most of the rhythmic layering. Horizontal currents along the bottom of the chamber probably explain the igneous lamination (alignment of feldpars NOT the layering). When a pulse of crystallisation produces a large proportion of dense crystals the convection current may become unstable and collapse (like column collapse in volcanic eruptions). This produces turbidity currents of crystals and magma which flow down the edges of the magma chamber and then across the base towards the centre. These flows produce trough and cross-bedding and cause erosion and unconformities in the layering. The crystals within the flows will be deposited as a normally graded layer. 2.1.d. Cumulates Once crystals have settled to form a cumulate layer at the base of the magma chamber evolution does not stop. Adcumulus processes can continue to change the texture and composition of the rock. In the simplest case the intercumulus liquid trapped between the cumulate phases crystallises to form a poikolitic texture (often of pyroxene around plag) such rocks are known as orthocumulates. Adcumulus growth of the cumulate phases, however, can occur in which the cumulate minerals increase in size by growth from the intercumulus liquid (somewhat like an overgrowth cement during the diagenesis of clastic sediments). The resulting adcumulate can be monomineralic and the unwanted components in the intercumulus liquid must have been removed (mesocumulates are those in which adcumulus growth has not gone to completion). Because the intercumulus liquid is not necessarily in equilibrium with the cumulates it can react with one or more of the cumulate phases by adcumulus replacement. Compaction of the cumulate layer and adcumulus growth will force the intercumulus liquid to migrate away through the cumulate pile. This liquid may react with the cumulate it flows through to cause infiltration metasomatism. Intercumulus liquid may also collect and intrude through the cumulate layer. On a larger scale gabbroic rocks found in the lunar highlands may have been intercumulate magmas leftover from the crystallisation of the lunar mantle Reduction and Oxidation the role of oxygen fugacity. Free oxygen in a magma is enormously reactive because it has two spare electrons available to share with cations. Reactions between free oxygen and multivalent cations (cations with more than one valence state) increase the formal positive charge on the cations since it must share some of its electrons with the oxygen and thus becomes more positive. Iron cations are the commonest multivalent cations in geological systems and the abundance of iron can drastically change the identity of the iron-rich minerals crystallising from a magma and thus the effect of crystal fractionation on the composition of the magma. Consider the reaction of fayalite (iron-rich olivine) with free oxygen in a magma to produce magnetite and quartz. 3Fe 2+ 2SiO4 + O 2-2 = 2(Fe 2+ O.Fe 3+ 2O3) + SiO2 (reaction 1)

5 If we increase the oxygen fugacity (the activity of oxygen) then more fayalite will react to form magnetite and quartz. In this reaction we are increasing the amount of trivalent Fe 3+ (ferric iron, oxidation state III) compared to divalent iron Fe 2+ (ferrous iron, oxidation state II). A reaction that causes iron (or any other multivalent cation) to increase the number of electrons involved in bonding (i.e. increases the overall valence state) is known as an oxidation reaction. If we reduce the oxygen fugacity in reaction 1 then we ll consume magnetite and quartz and create fayalite and free oxygen. This reduces the overall valence (oxidation) state of Fe (creating more ferrous Fe 2+ ) and is known as a reduction reaction. Both oxidation and reduction are redox reactions. Reaction 1 occurs at specific conditions of fo2 (oxygen fugacity) and temperature in much the same way as any other reaction (remembering oxygen fugacity is closely related to the abundance of free oxygen). Under equilibrium the presence of the reactants and products in a system means we must be somewhere on the univarient reaction boundary. Essentially at any temperature the oxygen fugacity is fixed by the reaction. The reaction is then said to be an oxygen fugacity buffer. Several other reactions involving Fe and O act in the same way and are similarly oxygen buffers. Important ones include metallic iron-wustite (low oxygen fugacities) and magnetite+quartz-hematite (high oxygen fugacities). These are shown on the phase diagram. 2.2.a. Metal-bearing Basalts. An excellent example of how magma compositions can be influenced by redox reactions occurs occasionally around tree boles in basaltic lava flows. The wood has burnt to charcoal and the basalt adjacent to the tree bole often contains iron-nickel metal and highly Mg-rich silicates. The reason for this unsual assemblage is that carbon from the wood has reacted with oxygen in the magma to produce CO2 which escapes as a gas. This is an oxidation reaction (carbon is multivalent) and removes free oxygen from the magma. The removal of oxygen decreases the oxygen fugacity and, therefore, causes reduction. The valence state of iron atoms in the magma decreases because of the reduction becoming Fe 0 (metallic) rather than ferrous (Fe 2+ ). Metallic Fe cannot be incorporated into silicate minerals and is not particularly soluble in silicate magmas and thus separates to form an immiscible metallic liquid. The remaining silicate magma, having lost most of its iron, becomes Mg-rich. Similar reduction on a much larger scale occurs at Disko Island in western-central Greenland where basaltic lava flows intruded through coal seams and assimilated large amounts of reducing carbon. Reduction is also likely to be an important part of core formation in planets and melting of carbon-rich asteroids. 2.2.b. Oxygen fugacity in Igneous Rocks Oxygen fugacity frequently increases during fractionation of silicate magmas from basaltic to more acidic compositions. Often the ratio of Fe2O3/FeO (i.e. Fe 3+ /Fe 2+ ) can be used as an indicator of oxygen fugacity due to the effects of oxidation reactions on the valence state of Fe. Hematite, for example, which only contains ferric (Fe 3+ ) iron occurs in rhyolitic rocks as well as amphiboles that can incorporate Fe 3+. Continental Mantle-Derived Magmas Mantle-derived magmas erupt in continental areas in two main tectonic settings: (1) continental rifts, and (2) stable cratons. 1. Continental Rift Magmatism Continental rifts are linear zones of extension within the continental crust caused by mantle convection. Some of these extensional zones become zones along which the continents break apart to form new oceanic basins, however, in most cases such break-ups have failed. There are a number of distinct phases in the development of continental rifts related to different styles of magmatism. Rift development can stop at any stage Pre-rift Doming Stage Continental rifts initiate when a mantle plume (diapir) rises below the continental crust. The arrival of the mantle plume causes uplift and doming of the crust and thinning of the lithosphere by ductile flow. Once the plume rises into the lithospheric mantle large scale melting can occur generating mainly tholeiitic basaltic magmas with minor alkali basalt. They are erupted at the surface in large volumes. These are known as the continental flood basalts and can be extruded at rates of up to 1x10 6 m s -1 (see volcanism notes lecture 9). Flood basalt magmas probably accumulate in large magma chambers towards the base of the crust since they tend to have lower Mg contents (5-8%) than would be expected from primary mantle magmas and have probably undergone crystal fractionation. The are

6 generally poor in phenocrysts. Continental flood basalts have incompatible element patterns similar to EMORB indicating an asthenospheric mantle plume contribution. Their 86 Sr/ Sr and 143 Nd/ 144 Nd ratios overlap with continental crust suggesting some assimilation of the lower crust during residence within magma chambers Rift Stage Further extension leads to the development of grabens by faulting resulting in the formation of rift valleys. Initally these form as triple junctions (or rrr junctions) above the rising plume. Three distinct types of magmatism are observed within rift systems centred along the rift axis. Alkali basaltic magmas that erupt to form shield volcanoes or monogenetic (single stage) scoria/cinder cones. Often scoria cones are located along transform fault traces. The large shield volcanoes are often developed on the margins of the rift system. Alkaline magmas (phonolites, nephelinites etc) erupt towards the rift axis forming stratovolcanoes. Rhyolites magmatism is also located in the rift and is particularly common earlier in rifting. The origin and distribution of magmatism in the rift stage relates to the depth of magma generation and degree of partial melting. Alkaline magmas are probably generated at deeper levels in the hot rising plume and evolve through crystal fractionation to form highly alkaline phonolites and nephelinites or alkali-rich rhyolites (see below). Alkaline magmas frequently contain megacrysts (single crystals) of mantlederived garnet suggesting a deep source. Basaltic magmas form either at shallower depths below the rift or due to higher degrees of partial melting further from the rift axis. The incompatible element systematics of rift alkaline magmas are usually enriched in incompatibles relative to EMORB suggesting a larger plume derived component. Some also have 86 Sr/ Sr and 143 Nd/ 144 Nd that overlap with continental crust indicating assimilation Afar Stage Once thinning of the lithosphere has brought the asthenosphere to crustal depths tholeiite magmas start to be erupted at the surface and the generation of oceanic crust occurs in the central region of the rift. Alkali basalts, generated at greater depth are erupted further from the rift axis. Further extension of the rift will produce oceanic crust and MORB-like tholeiites East African Rift The East African Rift which extends from Syria in the north to Mozambique in the south has been active throughout the Cenozoic. During the initial stages of rifting fissure eruptions produced large volumes of basalt and siliceous ignimbrites. During the late Miocene and Pliocene these eruptions became more focused, and produced shield volcanoes consisting of basanites, rhyolites and phonolites. In Plio-Pleistocene times rhyolites were erupted along the main axis of the rift, while basalts continued to be erupted on the plateaus adjacent to the rift. Quaternary volcanoes along the axis of the central rift zones, in Kenya and Tanzania, consist of phonolite, trachyte, or peralkaline rhyolite. This province illustrates the wide variety of unusual rock types found in continental rifting settings. Note, however, that parts of the rift along the Red Sea and Gulf of Aden have evolved to oceanic ridges and produce MORBs to form new seafloor Crystal Fractionation of Alkaline Magmas: The Residual System Small degrees of partial melting at depth generate magma derived silicaundersaturated alkaline magmas (see lecture 12). On crystal fractionation during ascent or residence in a magma chamber these can produce either rhyolites or phonolites etc. The reason these magmas produce two fractionation trends is a partial thermal divide exists at low pressure once these magmas have evolved by crystal fractionation to trachytes and are crystallising alkali feldspar. If they are silica-undersaturated fractionation of alkali feldspar produces Na+K rich magmas. If they are slightly silica-saturated (on the silica side of the thermal divide) they become more silica-rich by fractionation of alkali feldspar and evolve to rhyolite magmas. Assimilation of silica-rich crustal rocks may drive alkaline magmas towards rhyolites. The thermal divide is shown in the system Nepheline-Silica-Kalsilite. 2. Cratons and Kimberlites

7 Kimberlites are volatile-rich potassic ultrabasic rocks found within conical pipes generally within stable cratons (i.e. thick stable crust). They are hybrids containing large amounts of xenolithic and xenocrystic material including olivine, enstatite, Cr-diopside, phlogopite, garnet and ilmenite that are fragments of mantle-wall rocks. Their matrices are finegrained and consist of serpentine, phlogopite, carbonates, perovskite and other minerals. Kimberlites often contain diamond as a trace accessory mineral Occurrence & Eruption Kimberlites (and related volatile-rich ultrabasic rocks, lamproites and minettes) are found in narrow, funnel-shaped bodies known as diatremes. Kimberlite within diatreme is fragmental and often contains large blocks of country rocks. These were emplaced during highly explosive eruptions driven by the high volatile-content of the kimberlite magma. At depth diatremes are connected to non-fragmental kimberlite sills and dykes. Kimberlites diatremes are often found in swarms and are connected at depth. Kimberlites are generally associated with stable cratons such as in S. Africa and Lesotho and Siberia, however, some are found marginal to cratons and continental rifting settings (i.e. Tanzania further from the rift) Compositions Two types of kimberlite are identified on the basis of their compositions and mineralogies. Group I kimberlites (or basaltic kimberlites) are CO2-rich and have Sr and Nd isotopes similar to fertile mantle. Group II kimberlites (or orangites) are H2O-rich and contain much phlogopite and mica, they have Sr and Nd isotopes indicating derivation from an enriched (but old) mantle source. Both types of kimberlite are enriched in incompatible elements relative to OIB. The origin of kimberlitic magmas are thought to be related to decreases in the solidus of peridotite at pressures >30kb in the presence of water and CO2 resulting in a phlogopite and carbonate bearing mantle peridotite source. Group I kimberlites are thought to have been derived from asthenospheric mantle (due to their fertile character) and contain fertile-sheared mantle. Group II kimberlites are thought to be derived by melting of lithospheric mantle that has been enriched by mantle metasomatism. The timing of kimberlite emplacement suggests that partial melting occurs due to proximity to an asthenospheric mantle plume.