1. Partial melting of amphibolites and the genesis of Archaean TTG (and some geodynamical implications) Jean-François Moyen and Gary Stevens Stellenbosch University, South Africa

2. TTG are... Orthogneisses Tonalites, Trondhjemites & Granodiorites (Na-rich series) Fractionnated REE, etc. Largely homogeneous throughout the Archaean Originated by partial melting of amphibolites (hydrated basalts), in garnet stability field

3. Trace elements features of Archaean TTGs Nb-Ta anomaly Sr contents Y & HREE depletion

5. Les « gneiss gris »

6. Minéralogie

7. Eléments majeurs

8. REE

9. Conditions for making TTGs Experimental melts In Garnet stability field (Gt in residue) Melting of hydrous basalt KDKD Gt/melt = 10 - 20 (other minerals ≤ 1) Yb

10. Geodynamic site ? Thick (oceanic or continental) crust (e.g. Oceanic plateau) Subduction Intermediate cases: Shallow subduction (± underplating) Stacked oceanic crust Gt-in

11. Partial melting of amphibolites 15-20 « modern » studies (1990-2000) + Phase diagrams (1970-80) 114 exp. fluid present or saturated 209 exp. « dehydration melting »

12. Goal of the study Review and compilation of published data on experimental melting Elaboration of a global model for amphibolite melting Implications for trace element contents Geological/geodynamical consequences

13. Review and compilation of published work Starting materials Solidus position & melt productivity Mineral stability fields (Moyen & Stevens, subm. to AGU monographs)

14. Starting materials

15. Types de fusion ? 1) Fusion en présence d ’eau Minéraux + H 2 O -> Liquide Exple : Qz+KF+Pg+H 2 O -> Liq Pg+Qz+Amp+H 2 O -> Liq + ? 2) Fusion-déshydratation (« dehydration melting ») Minéraux hydratés -> Minéraux anhydres + H 2 0 Minéraux + H 2 O -> Liquide Exple : Bt+Pg+Qz -> Liq + (Opx+Crd)/Gt Pg+Amp -> Liq + Opx/Gt Min. hydratés -> Min. anhydres + liquide

16. Fluids and melting Fluid-saturated (free fluid phase) Fluid-present (yielded by breakdown of hydrous minerals in the near sub-solidus), limited availability Fluid-absent (dehydration melting) Dry

17. Fluid saturated

18. Dehydration melting

19. Fluid-present

20. Experimental solidus position

21. Melt productivity: dehydration melting

22. Melt productivity: water saturated (+ Qz)

23. Melt productivity: fluid- present (- Qz)

24. Mineral stability limits

25. Control on amphibole stability

26. Control on plagioclase stability

27. Mineralogical models CaO 11.010.09.0 Na 2 O 2.22.83.3 K2OK2O 0.10.51.0 TiO 2 1.22.10.8 Amp. Comp. Ti-rich High Mg# Si poor Int. Ti-poor Low Mg# Si rich KoBThBAB Quartz 0110 Plagioclase 254054 Amphibole 755936 Amp:Plag 3:13:22:3

28. Mineralogical models KoB ThBKoBThBAB

29. Composition of experimental melts

30. Very unlikely for amphibolite melting! Na 2 O contents in experimental melts

31. K2OK2O

32. Major elemen ts A linear model, of the form C/C 0 = a F + b

33. Modelled melts

34. Model vs. TTGs

36. Preliminary conclusions (1) K2O content depends on the source. Only relatively K-poor sources (< 0.7 %) make TTGs … but really depleted sources won’t. This means that K-rich amphibolites can indeed melt into granites (Sisson et al., 2005) With appropriate sources, tonalites & trondjhemites occur for F = 20-40 % (900- 1100 °C)

37. Model for trace element ClCl = C0C0 F + D (1 - F) Experimental data D =  Kd i. X i Arbitrary Litterature

38. KoB ThBKoBThBAB Trace elements contents of the 3 sources

39. Melt proportions KoBThBAB

40. Mineral proportions: amphibole and plagioclase KoBThBAB

41. Mineral proportions: garnet KoBThBAB KDKD Gt/melt = 10 - 20 Yb

42. Mineral proportions: rutile KoBThBAB KDKD Rt/melt = 25 - 150 Nb KDKD Rt/melt = 50 - 200 Ta

43. REE contents in (modelled) melts KoBThBAB

44. REE contents in (modelled) melts KoBThBAB

45. REE contents: La/Yb KoBThBAB

46. Y contents KoBThBAB

47. Sr contents and the role of residual plagioclase (Martin & Moyen, 2001, Geology 30 p 319-322; after Zamora, 2000)

48. Sr/Y KoBThBAB

49. Nb/Ta KoBThBAB

50. Effect of pressure

53. TTG composition as a depth indicator Nb-Ta anomaly and Nb/Ta Sr contents Y & HREE depletion

54. TTG composition as a depth indicator (cont.) HREE depletion Eu anomaly

55. Preliminary conclusions (2) Appropriate depletion in Y, Yb, etc. requires pressures above ca. 15 kbar (rather than 10 kbar = Gt-in) Y, Yb, Sr/Y, Nb/Ta etc. are indicators of melting depth Low- and high-pressure TTGs with contrasted signatures?

56. High P TTGs Low P TTGs Not really TTGs Archaean granulites (and intraplate geotherms) Subduction of young lithosphere (5 Ma) (20 Ma) (50 Ma) Subduction of old lithosphere Tonalites & trondhjemites (F = 20-40 %) Appropriate trace elts. signature High Sr, La/Yb, Nb/Ta Low Y, Yb Low Sr, La/Yb, Nb/Ta High Y, Yb TTG genesis in P-T space

57. A regional example Barberton, South Africa 3.5 to 3.2 greenstone belt and gneisses

58. Swaziland R.S.A. 20 km Crust accretion around BSB 3600-3500 Ma Steynsdorp pluton 3509 ± 7 Ma Ngwane gneisses (Swaziland) 3490 ± 3 to 3644 ± 2 Ma Lower Onverwacht group ca. 3500 Ma Dwalile Suite greenstone remnants Ca. 3500 Ma ?

59. Swaziland R.S.A. 20 km Crust accretion around BSB 3450 Ma Stolzburg, Theespruit, etc. plutons 3443 ± 4 to 3460 ± 5 Ma Tsawela gneisses (Swaziland) 3458 ± 6 to 3437 ± 6 Ma Upper Onverwacht group ca. 3400 Ma

60. Swaziland R.S.A. 20 km Crust accretion around BSB 3220 Ma Kaap Valley, Neelshoogte, Badplaas, etc. plutons ca. 3220 Ma Usutu granodiorite (Swaziland) 3231 ± 4 to 3216 ± 3 Ma Fig Tree and Moodies groups ca. 3200 Ma Dalmein pluton Ca 3220 Ma

61. Crust accretion around BSB 3100 Ma & younger Swaziland R.S.A. 20 km GMS suite: Piggs Peak, Heerenveen & Mpuluzi batholiths Boesmanskop syenite ca. 3107 Ma Ushuswana complex 2900 Ma & K granites 2700-2650 Ma

62. Geochemistry: 3600-3500 Ma Steynsdorp pluton Ngwane gneisses

63. Geochemistry: 3450 Ma event Stolzburg & Theespruit plutons Tsawela gneisses

64. Geochemistry: 3220 Ma event Kaap Valley, Nelshoogte & Badplaas plutons

65. TTG evolution around Barberton Greenstone Belt 3.6 – 3.4 Ga 3.4 – 3.2 Ga

66. Amphibolites with HP relicts

68. Preliminary conclusions (3) TTGs in Barberton record progressively deeper sources This is consistent with progressive steepening or onset of subduction, and could witness the progressive accretion of a continental nucleus and its early growth At 3.2 Ga (true subduction established), the geothermal gradient recorded in some metamorphic rocks is consistent with the gradient corresponding to TTG genesis

69. Secular/Geodynamical implications Progressively cooler gradients ? Early Archaean Late Archaean Modern

70. Geodynamical implications Steepening/onset of subduction ?

71. Preliminary conclusions (4) Secular chemical evolution of TTGs reflects increasing melting depth and increasing interactions with the mantle This is consistent with a subduction origin for TTGs Secular cooling of the Earth makes the melting deeper and deeper along the subducted slab, allowing more and more interactions with the mantle Alternately, this could witness progressive onset of subduction

72. Conclusions TTGs are diverse, and their chemistry reflects the depth of melting; melting occurred mostly at 15-20 kbar, but can have occurred anywhere between 10-12 and 30 kbar. Most TTGs are probably originated in subductions, and interacted with the mantle to some degree The changes in TTG compositions can probably be correlated with changes in tectonic styles – either in terms of secular evolution, or in one single area

73. The Sand River Gneisses Ca. 3.1 Ga TTG gneisses in Messina area, Limpopo Belt, South Africa (R. White, Melbourne, for scale)