Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen
This course looks at the geodynamic and tectonic context for the formation of continents.
Petrogenetic mechanisms proper (i.e. petrogenesis of crustal rocks) will be discussed in the lectures by Hervé Martin.
Continental crust accretion:
In volume, it occurs mostly in the Archaean (up to 80% of present CC volume).
On a regional basis, at the end of the Archaean cratons are formed and are stable continental masses, that can support large intracratonic basins and have a lithospheric root; the cold, thick, resistant lithospheric roots is what makes a craton: a stable block of continent that can survive for long periods of time, and is only marginbally affected by orogenies etc.
Geochronology suggests that the cratonic root was acquired also at the end of the Archaean, so crust and mantle form more or less at the same time.
The question: how do you build a continent (during the Archaean), in terms of geological/tectonic processes?
Three lines of evidence:
Geological structures. Map pattern ,deformation style, and in general geology of Archaean
cratons is clearly not “normal” (see part 0)
Physical constrains. Earth’s mantle heat mostly comes from radioactive decay of U, Th, K and therefore decreases exponentially with time. Archaean heat production was 2—3 times higher.
Petrological constrains. For primary mantle melts, there is a correlation between [MgO] and melting temperatures. Geological evidence show that T decreased with time, although the exact details of the T-time relations are discussed (regular decrease ? Stepwise? Peak temperature then decrease?)
fact, PT is a very special mode of convection, that involves a conductive thermal boundary layer (the lithosphere) being broken and forming the down-welling branches of the convection system.
The mantle of telluric planets is convecting. However, not all planets do have plate tectonics. In
1

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen
A much more “logical” mode of convection would be “stagnant lid”, with an immobile, intact TBL floating on top of the convection mantle.
From a surface perspective, plates are moved primarily by slab pull (dragged by the weight of the slab sinking in the mantle).
The naïve view: temperature of Earth’s mantle follows heat production. This does not work because the temperature is actually a balance between heat production and heat loss. Heat loss is controlled by the efficiency of the convection/global tectonics system, in turn a function of mantle T: there is a potentially complex feedback loop.
Present day Earth looses heat mostly through oceanic basins, at ridges and young oceanic lithosphere (thin, less resistant to heat transfer). Heat flux from the continents is nearly matched by heat production in continents, does not come from the mantle and plays no role in convection systems. Total non-continental heat loss is ca. 36 TW.
Heat production: radioactive decay of U Th K. Cosmochemistry (chondritic models etc.) suggests total heat production of ca. 13 TW. Urey ratio:
U
R

H
Q blanace between heat loss and heat production; present day U
R
≈ 0.2 to 0.4.
Conventional scaling of tectonic processes suggests heat dissipation is proportional (or positively correlated) to mantle temperature (largely because high temperature  low viscosity  faster convection?). This implies that Earth was cooling even faster in the past than now, mantle temperature decreased exponentially; extrapolating back to the Archaean predicts extremely high mantle temperatures (exceeding 2000 °C at 2 Ga): “thermal catastrophe”
Possible resolutions of the heat paradox:
U
R
is higher than suggested by geochemistry; with U
R
≈ 0.7 it is possible to produce a sensible thermal history. o Heat production is higher:
 Cosmochemical daa not reliable, mantle richer in U Th K;
 Some heat is supplied from the lower mantle or the core; o Heat loss is lower: present day Q is abnormally high (breakup of Pangea: too many ridges)
Conventional scaling is wrong, heat dissipation is weakly or negatively correlated to mantle T.
In this case one can produce thermal histories with T increasing from 4 to 3 Ga, and decreasing after that. This matches (some) models derived from lava [MgO]. This implies an inefficient convection in a hotter Earth.
Three main effects can play a role:
Thicker oceanic plates. Hottter mantle implies more melting at ridges, and therefore both thicker oceanic crust and thicker harzburgitic lithosphere. Two effects:
2

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen o Thicker plates  smaller heat diffusivity, less conductive heat loss through oceans; o Both oceanic crust and harzburgites are buoyant  subduction initiation more difficult (on the other hand thick oceanic crust transforms into eclogites, which are dense; once subduction is initiated this may counterbalance the other effect)  less subduction, lesser slab pull, less mobile plates, older lithosphere in average, less heat loss.
Weak subductions in hotter mantle. Subducted lithosphere is heated and becomes less resistant, breaks off easily, transition to “drip” regime rather than proper subduction.
Balance between decreasing viscosity  faster plates and temperature  weak plates, apparently the subduction “efficieny” peaks at mantle temperatures ca. 100 °C higher than today and decreases then. This allows to predict thermal histories with a peak at ca. 3 Ga, similar to scaling models.
Difficult subduction initiation in hotter mantle, weak plates buckle and do not break.
Transition to “stagnant lid” regimes. If the TBL (lithosphere) is too strong, it does not break and convection operates as stagnant lid; if mantle too hot, lithosphere buckles and does not break either. Stagnant lid regimes would exist for strong (thick?) lithosphere and/or hot mantle. Geology of Venus may suggest stagnant tectonics with coronas (trace of upwelling zones?) and “tessera terrain” (traces of downwelling regions and compression of the nonfailed lithosphere ?).
Note that “subduction” does not imply similar to modern. Modern subduction is one-sided (only one side of the contact goes down), two-sided subduction is plausible too. Subduction is one sided if the cold slab is stronger than the hotter backarc, and fluids help in decoupling the slab from the wedge.
Metamorphic records: two types of geothermal gradients co-exist since at least 3.0 Ga (sparse data before), suggesting one-sided subduction is operating since then.
Both mechanical models and heat flux balance considerations suggest a less efficient
Archaean convection/tectonic system
Archaean subductions (=drag for plates), and therefore plate tectonics, were probably o More difficult to initiate o With more slab breakoff o Possibly, two-sided
Archaean oceanic plates may have been thicker
Can we imagine the geology of Earth without subduction? Dripping of the lower crust into the mantle? Melting in the crust, either above drips (heating by crustal root removal) or above rising zones (oceanic plateau-like melting?)
Evidence for Archaean plate tectonics? We must look for convergent plate boundaries (where most of the geology happens): structures, metamorphism, magmatism are typical from convergent boundaries in the modern Earth. What can we find in the Archaean?
3

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen
In this section we’ll be looking at a few, relatively well-studied, examples of Archaean cratons:
The Superior Province in Canada, and mostly the Abitibi sub-province (2.7 Ga, with herited ages up to 3.1 Ga)
The Kaapvaal craton, in South Africa (3.5 – 2.7 Ga);
The Pilbara craton, in Australia (3.5 – 2.8 Ga);
The Dharwar craton, in South India (mostly 2.5 Ga for the studied features, older ages extending to 3.0 Ga);
The Yilgarn craton in Australia (mostly 2.7 Ga).
Archaean cratons are typified by three geological units:
Greenstone belts, i.e. associations of supracrustal rocks (lavas and sediments). Lavas are mostly mafic and ultramafic (komatiites) lavas.
Plutons, typically syn to late tectonic; they can cover a range of rock types (discussed further
in section 3.1.2).
Grey gneisses, the “background” lithology of the cratons; a complex unit made of tectonically interleaved plutonic rocks and supracrustals.
The map pattern of most Archaean cratons is typified by domes of granitic/gneissic rocks surrounded by narrow “bands” of greenstones;
Greenstones typically form complex synforms, although in details it is rare to have a proper, symmetric syncline – mostly different units stacked together in a synform.
The granite/gneiss domes are said to display a concentric foliation pattern, this is however not always obvious in details…
Gravimetry allows to “see” the structures at depth; typically the greenstones extend down to
5—6 km and have a flat bottom (detachment between greenstones and middle crust?).
The domes can actually be a range of different things: o Syn-tectonic plutons o Migmatitic domes o Complex units of previously structured crust
Metamorphic studies reveal that gneissic domes are almost always at mid-crust PT conditions, whereas the greenstones are greenschist to sub-greenschist.
The contact between greenstones and domes is mostly tectonic (except when the domes are made of intrusive plutons), with a “greenstone down” sense of shear.
2.3.1.
Deformation styles
4

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen
In the greenstones: tight synclines and longitudinal faults; inconsistent stratigraphy between parts of the synform. Mostly low strain in the greenstone. Probable basal detachment. In many cases, relative movement with the gneisses (greenstone down), occasionally radial sinking.
In the mid-crust gneisses: bulk homogeneous shortening accommodated by vertical foliations, conjugate shear zones.
No evidence for significant crustal thickening: lack of thrust, thrust-and-fold belts, etc.
2.3.2.
Physical constrains on rheology
Hotter mantle implies higher heat flux in the crust, higher geothermal gradient in the crust.
Consequences:
Thin-skinned brittle upper crust, overlying a tick, weak middle crust (and below? Not sure, different possibilities: o Strong lower crust (melt depleted granulites) o Strong lithospheric mantle o No strong layer (“no lithosphere”) – a bit unlikely, would imply extreme geotherms.
No thickening possible – lower/middle crust flows laterally and does not allow thick crust.
Role of volume forces, gravity-driven tectonic. Especially since the upper crust (basalts) in denser than the middle/lower crust (gneisses); in a hot, low-viscosity crust, it is possible to
“sink” the upper crust in the lower crust.
At the continent level, Archaean cratons appear as the association of two types of structural elements:
Domes (more or less elongated)
Bulk crustal shortening with lateral extension, that can be localized and partitioned into a network of shear zones (conjugate, rotated into parallel), depending on the crust rheology
(weak crust = more homogeneous shortening, stronger crust = more partitioning).
Interference between the two strain fields, resulting from bulk shortening and from graviy-driven tectonics. Results in a range of map patterns from circular to elongate, depending on which of the component dominates locally.
However, seismic data are problematic: they mostly show horizontal reflectors, hard to reconcile?
All models should account for all the observations above. Many “tectonic” models propose scenarios exactly similar to modern orogenies (thrusting etc.), which seems at odds with observations and physics.
Modes that take into account the specifities of Archaean deformation propose the following interpretations for the Archaean strain patterns:
5

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen
2.5.1.
Non-mobile models
Purely gravity-driven models: in-situ emplacement of thick greenstone sequences (basalts + komatiites) creates gravity instability + thermal blanketing allowing the crust to warm  “partial convective overturn” of the crust, with sinking greenstones and rising domes.
Has been proposed in the Pilbara craton (primarily), Northern superior (Minto), Western
Dharwar craton, Zimbabwe (not recently) – all regions with a very “ovoid” map pattern and not many shear zones or structural “grain”.
This model strongly implies a stagnant lid (or drip tectonics sometimes) convection mode; shortening could be related to local effects in a stagnant lid above downwelling regions of the mantle
(cf. tessera terrain of Venus).
Plutonic rocks in this model could be formed by partial melting at the base of the mafic pile – seems difficult to reconcile with geochemical data, unless dripping of the lower crust “mimics” a subduction.
The model seems difficult to reconcile with the clearly linear map patterns and intense
shortening observed in some of the cratons (see 2.6).
2.5.2.
Extensional models
Doming is initiated by extension (similar to modern core complexes); gravity-driven tectonics and buoyancy, or in some models subsequent compression, steepens the fabrics and forms the steep domes.
Model matches well the shallow nature of greenstone belts, and some interpretation of seismic data. More difficult to reconcile with the dominantly vertical surface fabrics, and the craton scale strain patterns. Also requires complex tectonic histories with several successive deformation phases, not always supported by geochronology etc. – rather, progressive deformation in a shot period of time.
2.5.3.
Convergence models
Or rather, interplay between convergence and gravity-driven deformation.
Convergence in a hot, soft crust can result in two sets of structures:
Crustal scale folds; amplification of the core of the anticlines by buoyant rise of granitic rocks.
Consistent with analog model for shortening of weak crust (buckling of the thin-skinned upper crust).
Conjugate (and rotating) shear zones (as in the Dharwar craton).
This models allows for some form of horizontal mobility and plate tectonics.
In this model, horizontal seismic reflectors are interpreted as detachment at the base of the weak middle crust (either the moho, or the top of a melt-depleted, rigid granulitic lower crust).
6

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen
Following on the idea of interplay between convergence and gravity-driven tectonics, one may interpret the geology of many Archaean cratons as resulting from a special kind of orogeny, involving minimum crustal thickening but mostly docking of exotic blocks: accretionary orogens (they are in part similar to modern accretionary orogens, such as in active continental margins).
Strain fields, as has been discussed, are in agreement with convergence playing a role in the structural development of cratons formed by blocks with different histories; greenstone belts seem to occur on sutures between distinct blocks
Metamorphism in potential “suture” greenstones shows (1) relatively high pressure metamorphism; (2) two types of metamorphic conditions (paired metamorphic belts?)
Accretionary orogens resulting in the collage of many small continental fragments? Cf. Western
North America, Eastern Australia.
Map pattern reveals that at a craton’s scale, continents are made of individual blocks with distinct timing and geological pre-history, that have been accreted together during deformation events.
Seismic data suggests this map pattern corresponds to deeper units with variously dipping contacts
At a more local scale, the stratigraphy of the greenstone belts also shows that they are
This part will be largely discussed by Hervé Martin and is only briefly touched on here.
Archaean rocks are mostly igneous; sediments are rare (and do not mark clear tectonic processes, maybe with the exception of mature sandstones and conglomerates).
3.1.1.
Volcanic rocks
Mafic lavas: o The most common rock type: tholeitic basalt. Origin by melting of the mantle. o A minor type: komatiites. High degree ofmelting, come from melting of a mantle hotter than ambient mantle (hot spot?)
Intermediate/felsic lavas: o Differentiated tholeites o Uncommon examples of “arc” rocks including adakites, shoshonites, boninites, etc.
Evidence for subdution or subduction-like processes.
7

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen
3.1.2.
Plutonic rocks
Crustal granitoids o Potassic: reworking of older continental rocks for the most part (older TTG, maybe sediments especially volcaniclastic sediments from intermediate lavas) o Sodic: the TTG series. Partial melting of amphibolites, probably form in three pressure bands (little garnet, grt + plag, no plag and grt only)
 LP: melting at < 12 kbar, may correspond to oceanic plateaux etc.
 MP: Melting at 15 kbar, consistent with collisional scenario (PT similar to metamorphic records)
 HP: melting at 20+ kbar, probably subduction related.
Mantle-derived granitoids (at least in part) o Sanukitoids and related rocks. Diorites evolving to (or associated with) granites; the diorites are very enriched, form from a metasomatized mantle. Metasomatic agent not sure, fluids, sediments or melts from subducted basalts (= TTG melts). Probably form after a subduction or subduction-like event.
3.1.3.
Associations in time
4 main magmatic associations:
“plume” association: tholeites, komatiites, low-pressure TTG. Can form in an oceanic plateau/mid-ocean ridge/etc.
“arc” association: high pressure TTGs, adakites. Seem to reflect subduction in some form (see discussion in part 1 however, could be any kind of burial of surface matter irrespective of the shape – even drips). Typically short lived (less than 10 Ma).
Syn-tectonic (“collision”?) association: MP TTG, crustal granites. Associated with peak deformation.
“post-collision” association: sanukitoids, shoshonites, crustal granites. Typically occur together with conglomerates and gold deposits. Probably remelting of previously metasomatized mantle, in “post-subduction” (or whatever it was) scenarios.
Lithosphere has a complex definition; it can be a thermal boundary layer (conductive zone), a mechanical behaviour, or a chemical reservoir. Below cratons, lithosphere extends to ca.
250—300 km commonly.
SCLM is known from inclusion suites in volcanic pipes (kimberlites).
The sub-continental lithospheric mantle of Archaean cratons is chemically complex, with a superposition of at least two successive events: o Intense melt depletion (up to 0 % melt removed); o Refertilization by silica addition and conversion of olivine to opx.
The SCLM is consequently light, as it is Fe-depleted and Grt and Cpx free; this compensates its cold geotherm and allows buoyancy.
Lithosphere also contains eclogite nodules; their composition is either (1) basaltic (oceanic crust?) or (2) complementary to TTG granitoids, i.e. either cumulates or restites. They could
8

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen correspond to some of the deep reflectors observed by seismic in the cratonic root. Dense eclogite at the base of cratonic root could help to stabilize the craton.
3.2.1.
Primary formation of SCLM
Depletion signature indicates up to 40% melt was removed from cratonic peridotites (much more than in phanerozoic peridotites, even oceanic).
Based on models of melting of fertile peridotites, melting occurred at < 3 GPa mostly.
This corresponds to potential temperatures of 1400 – 1500°C, 150° higher than modern
MORBs but far from what is inferred for komatiite plumes.
3.2.2.
Reworking of SCLM
Many evidence reveal that SCLM was modified after the initial depletion event:
Chemical data suggest melt extraction at < 3 GPa, even though peridotites record final equilibration at 3—7 GPa.
Deep seismic data could be interpreted as flat reflectors in the deep lithosphere (stacking of lithosphere slices?)
Metasomatism affects mostly deep peridotites: infiltration of siliceous melts or fluids from the bottom? No clear timing of metasomatic events. The nature of the metasomatic agent suggests fertilizing by originally shallow material (buried in subduction?).
3.2.3.
Coupled formation of crust and mantle
Events in the mantle seem to correlate more or less with events affecting the crust (but it is difficult to be positive, as there are many uncertainities around the dating of events in peridotites, nodules are not in-situ, etc.).
Very speculative model, 4 stages of cratonic formation:
1.
“oceanic” stage. Formation of a thick basaltic crust on “proto-ridges”. Plates are slow so no proper mid-ocean ridges, rather diffuse extension and volcanic centers. Local deep plumes produce komatiites, otherwise magmatsim is dominated by basalts. No continuous basaltic flow unit, rather overlapping volcanic centers. Residue of melting creates a primary lithosphere, depleted.
2.
Proto-continents. Modification of the thick basaltic lithosphere can occur in two settings, (i) in plume-related environments: proto-lithosphere is hit by plume, adding more mafic/ultramafic melts to the crust, and more lithosphere below; this also allows melting at the base of the crust to form granitoids, and probably associated diapiric deformation. (ii) above subductions or quasi-subductions (with slab breakoff and sinking of short bits of lithosphere), associated “arc” magmatism including deep TTG, boninites, adakites, etc.
9

Geodynamic of continental accretion
M2R Clermont-Ferrand – Nov. 2010 – J.-F. Moyen
Fertilization of SCLM. Depending on cratons, a crustal segment can see either or both scenarios.
3.
Collision between proto-continents. Main deformation event associated with limited burial of proto-continents, mostly transpressional accretion, crustal melting, remnants of arc magmatism, enrichment of SCLM.
4.
Relaxation post-collision. Pervasive crustal melting and formation of a granulitic crustal residue in the lower crust, diapiric exhumation of deep crust and final development of dome and basin pattern, melting of enriched SCLM to form sanukitoids, final depletion of
SCLM and stratification of the crust, leading to craton stabilization.
10