Accepted Manuscript

Petrogenesis of 3.15 Ga old Banasandra komatiites from the Dharwar craton, India: Implications for early mantle heterogeneity

J.M. Maya, Rajneesh Bhutani, S. Balakrishnan, S. Rajee Sandhya

PII: S1674-9871(16)30025-1 DOI: 10.1016/j.gsf.2016.03.007 Reference: GSF 443

To appear in: Geoscience Frontiers

Received Date: 5 November 2015 Revised Date: 1 March 2016 Accepted Date: 3 March 2016

Please cite this article as: Maya, J.M., Bhutani, R., Balakrishnan, S., Sandhya, S.R., Petrogenesis of 3.15 Ga old Banasandra komatiites from the Dharwar craton, India: Implications for early mantle heterogeneity, Geoscience Frontiers (2016), doi: 10.1016/j.gsf.2016.03.007.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

1 Petrogenesis of 3.15 Ga old Banasandra komatiites from the Dharwar

2 craton, India: Implications for early mantle heterogeneity

* 3 J. M. Maya, Rajneesh Bhutani , S. Balakrishnan, S. Rajee Sandhya

4 Department of Earth Sciences, Pondicherry University, Puducherry-605014, India

5

6 *Corresponding author. Phone: +919443636422;

7 E-mail address: [email protected]

8

9

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

10 Abstract 11 Spinifex-textured, magnesian (MgO >25 wt.%) komatiites from Mesoarchean 12 Banasandra greenstone belt of the Sargur Group in the Dharwar craton, India were 13 analysed for major and trace elements and 147,146 Sm-143,142 Nd systematics to constrain 14 age, petrogenesis and to understand the evolution of Archean mantle. Major and trace

15 element ratios such as CaO/Al 2O3, Al 2O3/TiO 2, Gd/Yb, La/Nb and Nb/Y suggest 16 aluminium undepleted to enriched compositional range for these komatiites. The depth of 17 melting is estimated to be varying from 120 to 240 km and trace-element modelling 18 indicates that the mantle source would have undergone multiple episodes of melting prior 19 to the generation of magmas parental to these komatiites. Ten samples of these komatiites 20 together with the published results of four samples from the same belt yield 147 Sm-143 Nd

21 isochron age of ca. 3.14 Ga with an initial εNd (t) value of +3.5. High precision 22 measurements of 142 Nd/ 144 Nd ratios were carried out for six komatiite samples along with 23 standards AMES and La Jolla. All results are within uncertainties of the terrestrial 24 samples. The absence of 142 Nd/ 144 Nd anomaly indicates that the source of these 25 komatiites formed after the extinction of 146 Sm, i.e. 4.3 Ga ago. In order to evolve to the ε 147 144 26 high Nd (t) value of + 3.5 by 3.14 Ga the time-integrated MANUSCRIPT ratio of Sm/ Nd should be 27 0.2178 at the minimum. This is higher than the ratios estimated, so far, for mantle during 28 that time. These results indicate at least two events of mantle differentiation starting with 29 the chondritic composition of the mantle. The first event occurred very early at ~4.53 Ga 30 to create a global early depleted reservoir with superchondritic Sm/Nd ratio. The source 31 of Isua greenstone rocks with positive 142 Nd anomaly was depleted during a second 32 differentiation within the life time of 146 Sm, i.e. prior to 4.46 Ga. The source mantle of 33 the Bansandra komatiite was a result of a differentiation event that occurred after the 34 extinction of the 146 Sm, i.e. at 4.3 Ga and prior to 3.14 Ga. Banasandra komatiites 35 therefore provide evidence for preservation of heterogeneities generated during mantle 36 differentiationACCEPTED at 4.3 Ga. 37 Keywords: Komatiite; Dharwar craton; Early mantle differentiation; 142 Nd/144 Nd; Sm-Nd 38 geochronology 39

2

ACCEPTED MANUSCRIPT

40 1. Introduction

41 Komatiites since their discovery have provided useful constraints on the nature of 42 Archean mantle (Viljoen and Viljoen, 1969a, b; Arndt, 1975; Nesbitt et al., 1979; 43 Blichert-Toft and Puchtel, 2010). There is now a broad framework on genesis of 44 komatiites, though some details vary from region to region. Particularly, there have been 45 contrasting views on the extent of melting required to generate komatiite magmas 46 (Takahashi and Scarfe, 1985; Rajamani et al., 1985; McKenzie and Bickle, 1988).

47 Komatiites from Archean greenstone belts particularly proved useful in providing 48 information on extent of depletion of incompatible elements in their source regions and 49 also inheritance from the older events of mantle differentiation. Touboul et al. (2012) 50 reported excess in 182 W from 2.8 Ga old Kostomuksha komatiites indicating preservation 51 of heterogeneities generated during the early differentiation events. On the other hand 52 Blichert-Toft and Puchtel (2010) showed remarkable consistence of initial Sm-Nd and 53 Lu-Hf isotope ratios in komatiites over a wide spatial scale. They also showed that the 54 initial ratios decreased with time indicating progressive mixing of depleted and enriched 55 reservoirs which were created during the early diffMANUSCRIPTerentiation of the bulk silicate earth. 56 We, here report new elemental and isotopic results of coupled 147,146 Sm-143,142 Nd 57 system from Banasandra komatiites of Sargur group of western Dharwar craton to 58 constrain the age and petrogenesis of these rocks. With the help of coupled 147,146 Sm- 59 143,142 Nd system we also demonstrate the preservation of early mantle heterogeneities up 60 to Mesoarchean.

61 2. Geological framework

62 2.1. Geology of Dharwar craton 63 The DharwarACCEPTED craton (Fig. 1) dominantly comprises 3.36–2.9 Ga old Tonalitic- 64 Trondhjemitic-Granodioritic (TTG) gneisses referred to as Peninsular gneisses, 65 greenstone belts and 2.6–2.5 Ga old K-rich granitoid rocks (Beckinsale et al., 1980; 66 Peucat et al., 1993; Balakrishnan et al., 1999; Jayananda et al., 2008; Chardon et al., 67 2011; Hokada et al., 2013; Mazumder and Eriksson, 2015). An increase in the grade of 3

ACCEPTED MANUSCRIPT

68 metamorphism from greenschist to granulite facies reported from the north to south of 69 the craton is interpreted to be the result of tilting of the craton northwards and post 70 Archean differential denudation, exposing deeper level of continental crust in the 71 south relative to the north (Chardon et al., 2008). Archean greenstone-granite terranes 72 are well exposed in northern block of the Dharwar craton which is considered to be 73 made up of eastern and western parts (Fig. 1). N–S trending, mylonitized zone 74 proximal to the eastern margin of the Chitradurga greenstone belt was suggested as the 75 boundary between western and eastern parts of Dharwar craton (Chadwick et al., 76 1992).

77 Western Dharwar craton is characterized by larger, 2.9–2.6 Ga old greenstone 78 belts (Anil Kumar et al., 1996; Nutman et al., 1996; Trendall et al., 1997), belonging to 79 the Dharwar Supergroup and smaller, 3–3.58 Ga old Sargur group of greenstone belts 80 (Nutman et al., 1992; Peucat et al., 1995; Jayananda et al., 2008) mostly occurring as 81 variably sized enclaves within Peninsular gneisses (Fig. 1). Greenstone belts of the 82 Dharwar Supergroup consist of relatively subordinate quantities of mafic-felsic volcanic 83 rocks, voluminous pile of sediments with conglomeraMANUSCRIPTte unit at their base. The Sargur 84 greenstone belts are characterized by ultramafic-ma fic rocks, including komatiites, minor 85 felsic volcanic rocks, quartzites and meta-pelitic and carbonate sediments with rare barite 86 bands and the prominent ones are Ghattihosahalli, Jayachamarajapura, Banasandra, 87 Kalyadi, and Nuggihalli (Jayananda et al., 2008) (Fig. 1).

88 Komatiites in the Dharwar craton were first identified by Viswanathan (1974) in 89 Kolar greenstone belt and were studied in detail by Rajamani et al. (1985) and 90 Balakrishnan et al. (1990). Geological setting, structural and geochemical aspects of 91 komatiites of Sargur group were reported by Radhakrishna and Sreenivasaiah (1974), 92 Viswanatha et al. (1977); Seshadri et al. (1981); Srikantia and Bose (1985); Jaffri et al. 93 (1997) and ChardonACCEPTED et al. (2008). Jayananda et al. (2008) carried out geochemical and 94 Sm-Nd isotope studies on komatiites of Ghattihosahalli, J.C. Pura, Banasandra, Kalyadi 95 and Nuggihalli greenstone belts and reported whole-rock Sm-Nd isochron age of 3352 ± 96 119 Ma.

4

ACCEPTED MANUSCRIPT

97 2.2. Geology of Banasandra area

98 Sparse outcrops of komatiites were found around Banasandra, Birasandra, 99 Kodihalli and Kunikenahalli villages (Fig. 2), between N13°14.32 ′ to N13°17.77 ′ 100 latitudes and E76°39.57 ′ to E76°42.6 ′ longitudes and described in detail by Srikantia and 101 Bose (1985). Representative samples of komatiites were collected from locations 102 indicated in the Fig. 2.

103 Komatiites occur as elliptical, ~200 m wide bodies enveloped within talc-tremolite- 104 chlorite schist (Fig. 2) and are invariably serpentinized. Pillow structures are well 105 preserved at several localities such as to the east of Birasandra and to the south of 106 Kodihalli indicating formation of komatiites under subaqueous conditions. The pillows 107 are 20 to 60 cm long and 15 to 20 cm wide enclosed by cherty layer (Fig. 3b) which, at 108 places, are rich in magnetite formed by metamorphism of the ferruginous chert. 109 Carbonate veins, few mm thick, cut komatiites. Spinifex textured komatiites, 5 cm to 10 110 m thick, with typical radiating and randomly oriented sheaths of long bladed olivine and 111 pyroxene (Fig. 3a), mostly replaced by serpentine are found around Birasandra and west 112 of Kunikenahalli villages while at other places MANUSCRIPT komatiites occur as massive or nodular 113 serpentinites. South of Birasandra, spinifex textured units occur above the pillow 114 structured komatiites and at other places pillow structured komatiites show gradational 115 contact with massive and nodular types. Based on their occurrences it is inferred that 116 spinifex textured unit belongs to upper parts while pillowed, massive and nodular types 117 of komatiites formed successive lower layers of komatiite flow in the Banasandra area. 118 Preservation of spinifex texture and pillow structure in almost all the komatiite outcrops 119 is analogous to the komatiites reported from Onverwacht volcanic rocks of the Barberton 120 greenstone belt, South Africa (Arndt, 1975; Chavagnac, 2004).

121 The talc-tremolite-chlorite schist surrounding komatiites could represent highly 122 sheared and ACCEPTED deformed komatiites and ultramafic rocks which have mylonitized contact 123 with the surrounding rocks. Based on emplacement of komatiites under sub-marine 124 conditions and structural and deformational features associated with talc-tremolite- 125 chlorite schist, Sriknatia and Bose (1985) suggested that the komatiites were not spatially

5

ACCEPTED MANUSCRIPT

126 related to metabasalts of the Bababudan Group and were tectonically transposed over the 127 Peninsular Gneiss and Chitradurga greenstone belt.

128 2.3 Petrography of komatiites

129 The komatiites have been affected by seafloor hydrothermal alteration and upper 130 greenschist facies metamorphism but relict igneous textures are well preserved, though 131 they are largely pseudomorphed by secondary minerals. Skeletal and platy olivine and 132 bladed pyroxenes define the spinifex texture and about half of these are replaced by 133 mostly antigorite, iddingsite and in few samples by actinolite. The fine grained or glassy 134 matrix is replaced by finely interwoven mesh of serpentine and chlorite. Fine grained and 135 rarely euhedral magnetite occur particularly at the margins of blades and euhedral grains 136 of olivine as a result of metamorphic reaction where olivine alters to form serpentine and 137 magnetite as reported by Cheng and Kusky (2007) for the komatiites from west 138 Shandong, North China Craton. The massive and nodular komatiite units show euhedral 139 olivine surrounded by serpentine and chlorite dominated matrix. Euhedral chromite 140 grains and magnetite occur in minor quantity.MANUSCRIPT Highly sheared ultramafic rock that 141 envelops the komatiite bodies consist of mineral assemblage talc + chlorite + actinolite + 142 magnetite. The mineral assemblage found in Banasandra komatiites are similar to those 143 reported for the Abitibi and Barberton komatiites (Donaldson, 1982; Arndt, 1986; 144 Sproule, 2002). Tholeiite samples contain plagioclase, amphiboles and quartz as major 145 minerals. Some micro-spinifex textured komatiites contain narrow criss-cross olivine 146 blades in a matrix of mesh-textured serpentine (Fig. 3b). Opaque minerals like magnetite 147 and sulphide are found as accessories.

148 3. Analytical techniques

149 Eleven komatiite and two tholeiite samples from Banasandra, were analyzed for 150 major, trace ACCEPTED and rare earth elements. 146,147 Sm-142,143 Nd isotopic study was also carried 151 out on the same samples. As reference samples, besides the isotopic standards, three 152 komatiitic amphibolites (2700 Ma old) from Kolar greenstone belt in eastern Dharwar 153 craton (Balakrishnan et al., 1990) and a recent volcanic sample from Barren island,

6

ACCEPTED MANUSCRIPT

154 Andaman (Sheth et al., 2009) were also analysed with komatiites and tholeiites of 155 Banasandra. Samples weighing around 2 kg were crushed into small chips out of which 156 about 100 g of each sample was powdered to 75 m using vibrating-cup mill made of 157 agate. For analysing trace and major elements except silica, 0.5 g of each sample was 158 digested following the conventional acid-digestion procedure (Shapiro and Brannock,

159 1962) using HF-HNO 3-HCl acids in closed Teflon beakers on hotplate at 120 °C. 160 Elemental concentrations were determined using ICP-MS and ICP-AES available with us 161 at Pondicherry University. Silica was analyzed spectrophotometrically after sample was 162 fused with NaOH. USGS rock standards BCR-2, BHVO-2 and AGV-2 were also 163 analyzed along with the samples and the precision for major elements is < 2% and for 164 trace elements is about 5% (Appendix Tables A1–A3).

165 For isotopic analysis, an amount of 0.5 g of <75 m-size samples was taken for 166 digestion. Closed digestion of samples was carried out by keeping them in 14 mL vials at

167 ~140 °C for nearly 12 hrs after adding HF, HNO 3 and HCl in 7:3:1 ratio. A few drops of

168 HClO 4 were added to the mixture. On complete dissolution, it was split into two fractions 169 of which, one third was used to find out theMANUSCRIPT elemental concentrations using isotope- 170 dilution technique and the other part was used for isotope composition determination. 171 The fraction for isotopic dilution is mixed with a calibrated tracer solution which is 172 enriched in 152 Sm and 150 Nd isotopes. To ensure homogenization of tracer and sample, 173 the solution was completely dried and re-dissolved in 2N HCl. This solution was passed 174 through the Bio-Rad AG-50W-X8, 200–400 mesh column of 4.5 mL volume whereby all 175 elements other than REE were eluted. The REE collected as a group using 6N HCl was 176 passed through Teflon coated HDEHP columns to separate Sm and Nd from other rare 177 earth elements. These Sm, Nd fractions were then loaded on Re filaments and analyzed in 178 the Triton® (Thermo Finnigan) TIMS at Pondicherry University. 179 For measurement of 142 Nd, 0.5 g sample was digested using the procedure same 180 as above. AfterACCEPTED sample was completely dissolved, it was passed through 15 mL HCl 181 column filled with AG-MP-50 (100–200 mesh) resin to remove major elements like Mg, 182 Al, Fe, Ca, Mn and Ba. The residue was again passed through AG-50W-X8 (200–400 183 mesh) resin (chloride form) columns to separate REE from other elements. The collected 7

ACCEPTED MANUSCRIPT

184 REE fraction was passed through 5 mL α-HIBA ( α-Hydroxy Iso Butyric Acid) columns 185 filled with AG-50W-X8 resin to separate Nd from other elements like Ce and Sm which 186 cause isobaric interference. New resin was filled in these columns prior to the loading of

187 each batch of samples as the resin was of single use only. It was conditioned using HNO 3,

188 HCl, NH 4OH and α-HIBA, which was prepared gravimetrically and pH was maintained 189 at ≈ 4.5. Nd is collected using 0.25 M α-HIBA. The collected Nd fraction is again passed 190 through 3 mL HCl columns filled with AG-50W-X8 resin to get rid of any impurities that 191 might have remained. Pure Nd fraction is loaded on zone-refined Re filament which is 192 supported by an ionization filament and is analyzed using Triton® TIMS. The measured 193 data is corrected for mass fractionation using the exponential law (146 Nd/ 144 Nd= 0.7219). 194 140 Ce and 147 Sm were continuously monitored throughout the run. The contribution of 195 142 Ce and 144 Sm is found to be negligible and also confirmed by the absence of 196 correlation of 142 Nd/ 144 Nd with the intensity of Ce or Sm signal. Each run consisted of 27 197 blocks of 20 cycles per block. Before each block, baseline measurements were made for 198 30 s duration. Integration time for each measurement was 8 s and peak-centering, beam- 199 focus and amplifier rotations were performed between the blocks. International standards 200 like AMES and La Jolla were also analyzed withMANUSCRIPT each batch of samples. The external 142 144 201 precision for the ratio of Nd/ Nd of the standards is ± 3.3 ppm (1 σstdev ) ( n=18) during 202 the course of this study.

203 4. Results

204 4.1. Geochemistry

205 Major and trace element data of the samples are given in Table 1. MgO content of 206 Banasandra komatiites varies from 24.46 to 34.07 wt.%. The tholeiites analysed have 207 MgO contents of 5.19 and 5.22 wt.% (Fig. 4). Mg # was calculated as Mg/(Mg+Fe) in

208 mole fraction and varies from 0.81 to 0.92 for the komatiite samples. SiO 2 concentration 209 is low (28.74–44.76ACCEPTED wt.%) and LOI values range from 6.10 to 17.60. The two samples

210 with lowest SiO 2 concentrations B1/2 and B4/2A, of 28.74% and 30.03%, respectively

211 also has lowest concentrations of CaO making the CaO/Al 2O3 ratio <1. These same 212 samples have highest amount of total Fe. As would be discussed below, the trace-element

8

ACCEPTED MANUSCRIPT

213 concentrations of these two samples are also not consistent with the other samples and 214 therefore, we suspect these have been affected by alteration and/or metamorphism. When 215 plotted together with other komatiites reported from world over, the Banasandra samples t 216 plot almost in the middle of the array in Al 2O3, TiO 2, Fe 2O3 and CaO versus MgO plot 217 with decreasing trends (Fig. 4).

218 Jensen’s cation plot, based upon proportions of cations (Fe 2+ + Fe 3+ + Ti), Al and 219 Mg recalculated to 100%, is generally used to distinguish komatiites from other basalts. 220 The Banasandra samples fall in the komatiite field except for two samples, B1/2 and 221 B4/2A that plot in the field of komatiitic basalt. The two basalt samples plot at the 222 boundary between high-magnesium basalt, high Fe basalt, and calc-alkaline basalt (Fig. 223 5a). Results are also plotted in the CMAS diagram (O’Hara, 1968), which is based on the 224 experimental study estimating pressure of equilibriation of ultramafic magmas with 225 minerals of upper mantle using major element composition. This plot is made by

226 calculating the projection from diopside onto the plane of C 3A-M-S (Cox et al., 1984) 227 (Fig. 5b). Komatiite samples plot in the CMAS diagram between positions of melt 228 compositions generated at 4 and 6 GPa (TakahashiMANUSCRIPT and Scarfe, 1985; Takahashi, 1986), 229 which corresponds to 120–200 km in depth. The two peridotiitic komatiite samples plot 230 close to 8 GPa (250 km), suggesting that these komatiitic magmas were generated at 231 variable depths, between 4 to 8 GPa. Tholeiitic samples plot between 0 to 3 GPa but 232 closer to 0 GPa and magmas representing them could have formed at depths < 30 km 233 (Fig. 5b).

234 REEs in all komatiites and tholeiites were measured using ICP-AES and the data 235 are given in Table 2. Chondrite normalized (McDonough and Sun, 1995) REE plot shows 236 that Banasandra komatiites have LREE depleted to flat to slightly enriched HREE

237 patterns (Fig. 6a) which is consistent with other such komatiites with high Al 2O3/TiO 2 238 ratios (Blichert-toftACCEPTED et al., 2015) while both the tholeiites are LREE enriched and have 239 higher REE abundances (Fig. 6). The two samples, B1/2 and B4/2A, yielded REE 240 patterns with slight enrichment of LREE.

241 4.2. Geochronology

9

ACCEPTED MANUSCRIPT

242 147 Sm/144 Nd-143 Nd/144 Nd ratios (Table 3) of ten samples of komatiites from 243 Banasandra are plotted in isochron-diagram using Isoplot (Ludwig, 2001). The komatiites

244 yield an isochron-age of 3137 ± 190 Ma and initial εNd (t) of + 3.5 (Fig. 7) with a 245 considerable scatter reflecting in high MSWD of 65. Recently, Blichert-Toft et al. (2015) 246 reported similarly high scatter in the Lu-Hf and Sm-Nd isotope results yielded by fresh 247 drilled core samples of komatiites obtained under the ICDP program in Barberton 248 greenstone belt. They showed that regression taking different groups of whole-rock 249 samples, together with mineral separates, could not improve MSWD better than 20. They 250 reported Lu-Hf and Sm-Nd errorchron ages with MSWD of 118 and 42, respectively 251 which are close to the accepted zircon-age of the Komati formation. We could also 252 improve the scatter slightly (MSWD = 46) by including four samples of the same 253 formation reported earlier by Jayananda et al. (2008) in the linear array corresponding to

254 an age of 3139±150. Further, the initial εNd (t) for t= 3.14 Ga varies for individual

255 samples. One sample, that yielded an extreme value of εNd (t) is an outlier in the Sm-Nd 256 linear-array (errorchron) and excluded from the age calculation. This variation in the

257 initial εNd (t) values is also observed in other studies on komatiites (e.g. Blichert-Toft et al., 258 2015), and is not correlated with petrological, MANUSCRIPT geochemical or alteration indices. The 259 initial, yielded by errorchron is therefore, generally taken for discussing the petrogenetic 260 aspects (Blichert-Toft et al., 2015).

261 4.3. 142 Nd/ 144 Nd measurements

262 The 142 Nd/ 144 Nd ratios of komatiites from Banasandra were measured along with 263 terrestrial standards, AMES and La Jolla. Two tholeiites from Banasandra, three 264 komatiitic amphibolites from Kolar greenstone belt, and a basalt sample from active 265 volcano at Barren Island, Andaman were also analysed as reference samples. All the 266 samples were re-analysed to confirm that no systematic error has affected the results. In 267 total, eight samplesACCEPTED yielded high-precision (less than 5 ppm standard error) results which 268 are reported here (Table 3). All these samples do not show any correlations of 269 142 Nd/ 144 Nd ratios with the 140 Ce or 147 Sm intensities indicating that there was no mass- 270 interference from Ce or Sm in these ratios. The results of 142 Nd/144 Nd isotope study are

10

ACCEPTED MANUSCRIPT

271 shown in Fig. 8. All samples yield 142 Nd/ 144 Nd value similar to that of above two

272 standards (1.141874 ± 0.000003, 1σstdev).

273 5. Discussion

274 Results obtained during the present study are discussed to understand 275 petrogenesis, age of formation and implications for mantle differentiation in the early 276 history of the Earth. However, it is important to first establish that these samples preserve 277 primary elemental and isotopic signatures even after undergoing low-degree alteration 278 and metamorphism.

279 5.1. Effects of alteration and metamorphism on elemental mobility

280 It was noted very early in the studies of komatiites that “all komatiites are altered” 281 (Nesbitt et al., 1979). This statement is re-iterated in a recent study on drilled-core 282 samples from the best preserved komatiites in type locality of Komati formation of 283 Barberton komatiites by Blichert-Toft et al. (2015). They state: “No komatiites are 284 pristine – all have undergone hydrothermal alteration at or near the sea floor, followed by 285 metamorphism during accretion of the volcanic MANUSCRIPT sequence to the continent and associated 286 deformation.” However, numerous studies on komatiites also showed that alteration and 287 metamorphism does not affect all the elements equally, and generally REEs, including 288 Sm and Nd and HFSEs, are immobile and can provide important petrogenetic inferences. 289 The problem of alteration of komatiites and effect of alteration on the trace elements are 290 discussed in detail by Lahaye et al. (1995). It has been shown that effect of alteration 291 depends largely on the nature of alteration and fluid involved. Lahaye et al. (1995) found

292 that REE and even HFSE are affected more in alteration by CO 2-rich fluid compared to

293 H2O-rich fluid. The samples collected during the present study are mainly altered by low- 294 temperature hydration reactions as reflected by the mineral assemblage of 295 chlorite+serpentine+epidoteACCEPTED and by absence of any carbonatized minerals. Elemental 296 mobility during the aqueous alteration is assessed by the covariance of elemental 297 concentration with the loss on ignition (L.O.I.) as shown recently by van Acken et al. 298 (2016). We found that there are no systematic trends in elemental concentrations versus

11

ACCEPTED MANUSCRIPT

299 L.O.I for the REEs and HFSEs as also noted by van Acken et al. (2016). The usual 300 criterion to select the samples not affected by the alteration or metamorphism is to look 301 for consistency in the ratios of those elements which are generally known to be immobile 302 during alteration and metamorphism. It is also believed that if the magmatic character of 303 the samples is preserved, variations in major elements should be consistent with the 304 olivine fractionation, assuming olivine fractionation from a homogenized melt during 305 emplacement (Nesbitt et al., 1979; Lahaye et al., 1995). Most of the samples in present 306 study show consistent variations in major and certain trace element ratios except two. 307 While all the samples analyzed during this study plot in the same trend in variation 308 diagrams of major elements versus MgO as the other komatiites from around the world

309 (Fig. 4), samples B4/2A and B1/2, which have unusually high Al 2O3 and low CaO and 310 Mg #, fall away from these trends. REE patterns of these two samples are also unusual in a 311 way that they show depletion in Dy, Ho, Er compared to Gd and Lu (Fig. 6). Condie 312 (2003) have used ratios of incompatible elements such as Nb/Y, La/Nb, Zr/Y, Zr/Nb and 313 Nb/Th to show that these ratios record change in the nature of mantle sources of different 314 types of komatiites with time. He further concluded that at ~3.5 Ga the mantle, in 315 general, had depleted and recycled components. MANUSCRIPT We have plotted Nb/Y versus La/Nb 316 (Fig. 9) of Banasandra samples along with other komatiites as compiled by Condie 317 (2003). Banasandra samples except the two samples B4/2A and B1/2, show consistent 318 values indicating undisturbed ratios of trace elements. Furthermore, the slight negative

319 trend shown in the Gd/Yb versus Al 2O3/TiO 2 plot (Fig. 10) is consistent with 320 fractionation of high-pressure aluminous phase, such as garnet, either during melting or 321 crystallization. In this plot also samples B4/2A and B1/2 show higher Gd/Yb for their

322 Al 2O3/TiO 2 ratio compared to other samples. These two samples are also the only

323 komatiite samples that show negative f Sm/Nd indicating enrichment of Nd during 324 alteration. It, therefore, appears that except these two samples, other samples preserve 325 primary trace-elementalACCEPTED and isotopic signatures. Hence, these two samples are excluded 326 while discussing the petrogenesis.

327 The most important evidence to suggest preservation of primary elemental and 328 isotopic information is the Sm-Nd isochron diagram (Fig. 7). The linear array in the Sm-

12

ACCEPTED MANUSCRIPT

329 Nd isotope space is not due to binary mixing as we did not find any trend in the 330 143 Nd/ 144 Nd versus 1/Nd diagram (not shown). The linear array in the 143 Nd/ 144 Nd versus 331 147 Sm/ 144 Nd diagram therefore indicates the closed system behaviour of the samples for 332 Sm-Nd system. However, it can also not be asserted that Sm-Nd system was completely 333 undisturbed because there is a scatter more than the analytical precision of the 334 measurements of isotopes resulting in rather imprecise age and therefore, yields an 335 ‘errorchron’, in the strict sense. We therefore, propose that the low degree hydrothermal 336 alteration that affected these samples occurred soon after the emplacement and cannot be 337 resolved from the time of emplacement. Nevertheless, the mean initial isotope ratio, 338 derived from these samples can be used for understanding the petrogenesis and nature of 339 the mantle source.

340 5.2. Petrogenesis

341 Significant co-variance in major elements is observed for Al 2O3, FeO and TiO 2 342 which are negatively correlated with Mg #. The strong positive correlation of Ni with Mg #

343 is also observed indicating olivine control. The low CaO/Al 2O3 ratios (0.28–0.9) and high 344 Al 2O3/TiO 2 ratios (25–63.8) of Banasandra komatiites MANUSCRIPT indicate their similarity to the Al- 345 undepleted (AUK), Munro type of komatiites (Arndt et al., 1977) which are generally 346 believed to have formed by partial melting at depths where garnet was not a stable phase 347 in the residue. The values of Gd/Yb ratios are also consistent with garnet not being in the 348 residue (Fig. 10). However, a negative linear trend with gentle slope in Gd/Yb versus

349 Al 2O3/TiO 2 plot might indicate partial inheritance from a deeper source with komatiite 350 magmas representing different samples being result of different episodes of melting at

351 different depths. The progressive decreases in Gd/Yb ratio and increase in Al 2O3/TiO 2 352 implies progressive lowering of depth of melting (Fig. 10). A similar trend, as compiled 353 by Blichert-Toft et al. (2015), was observed by Robin-Popieul et al. (2012) in the 354 Barberton greenstoneACCEPTED belt. They used this trend to suggest that all three types of 355 komatiites: Al-depleted, Al-undepleted, and Al-enriched are present in the Barberton belt.

356 Condie (2003) has compiled various trace elemental data from the different 357 provinces of komatiites and other mantle rocks. Banasandra komatiites yield Nb/Y and

13

ACCEPTED MANUSCRIPT

358 La/Nb ratios that span some range of aluminium undepleted komatiites but are more like 359 aluminium enriched komatiites (AEK) (Fig. 9). Recently, Tushipokla and Jayananda 360 (2012) have studied adjoining Nagamangala komatiites and concluded that one third of 361 the samples fall into the aluminium depleted komatiites (ADK) while the other samples 362 are similar to AUK suggesting a similar wide range in certain trace element ratios. The 363 wide compositional range observed in the Banasandra and other Dharwar komatiites can 364 be explained by the model of sequential melting of previously depleted source. It was 365 proposed by Ohtani (1990) that these AEK could have been generated by re-melting of 366 the residue after separation of ADK with an assumption that ADK were generated earlier 367 in Archean than AUK or AEK. Robin-Popieul et al. (2012) also suggested a similar 368 model of re-melting of the refractory residue to explain the range of komatiites observed 369 in Barberton. This model of sequential melting was first proposed by Arndt (1975) and 370 discussed by Nesbitt et al. (1979) wherein it was shown that even though ADK could 371 have been alternatively explained by higher degree of melting at depths > 400 km leaving 372 garnet as residue, generation of AUK requires source to be melted repeatedly in order to

373 generate observed low ratios of CaO/Al 2O3 and high ratios of Al 2O3/TiO 2 relative to 374 chondrites and ADK. The present study of Banasandra MANUSCRIPT komatiites provides new data to 375 suggest that sequential melting model can explain petrogenesis of these komatiites. The 376 particular characteristics of interest, as discussed above, are the wide range of elemental

377 ratios yielded by komatiites of this belt. The samples that have high Al 2O3/TiO 2 have low 378 Gd/Yb and Sm/Nd ratios. These together with major element variations particularly the # 379 variations of CaO, and Al 2O3 with Mg indicate that multiple melting, probably at 380 different depths, together with fractionation of olivine from the melt at the time of 381 emplacement, has caused the spread in compositional range

382 Modelling other trace elements like Ni, Zr and Nd suggests 60 to > 80% of partial 383 melting with only olivine being residue (Fig. 11). But, such high degree of melting, 384 during a singleACCEPTED melting event, appears unlikely as the melt and residue would have 385 separated at much less than 20% of melting (McKenzie and Bickle, 1988). This problem, 386 however, was overcome by Robin-Popieul et al. (2012) by citing results of an 387 experimental study that showed that at very high pressures (~13 GPa), the komatiitic melt

14

ACCEPTED MANUSCRIPT

388 would have the same density as the residue and, therefore, would not be able to get 389 separated by rising up buoyantly. They proposed that the compositional range observed 390 in Barberton can be explained by the re-melting of the source on rising to shallower 391 depths. It is, therefore, possible that the komatiites represent magmas that were generated 392 from a refractory residue that already had undergone several melting events adding up to 393 as high as 80%. To test this hypothesis we modelled the Sm-Nd concentrations which 394 span too wide a range to be generated during one single event of fractionation. It is found 395 that just two episodes of partial melting as low as 10%, of a primitive mantle source (Sun 396 and McDonough, 1989) would give rise to a residue which is depleted enough to yield 397 the observed range of Sm/Nd ratio on further melting (Fig. 12). In this modelling the 398 residue is taken to be garnet free indicating that the melting is at pressures lower than 5 399 GPa (Takahashi and Scarfe, 1985) where garnet is not a stable phase. However, this does 400 not preclude the possibility of earlier episodes of melting at deeper depths leaving garnet 401 as residue, consistent with the suggestion that AUK are generated after extraction of 402 ADK in Archean (Ohtani, 1990; Robin-Popieul et al., 2012). We do see contribution of 403 the deeper melts in our samples in terms of slight negative trend of Gd/Yb versus 404 Al 2O3/TiO 2. We, therefore, propose that the MANUSCRIPT source of Banasandra komatiites has 405 undergone more than two episodes of melting starting from 250 km as illustrated in Fig.

406 13. This model explains the wide range seen in the Sm/Nd, Gd/Yb and Al 2O3/TiO 2 (Fig. 407 13).

408 5.2. Age of Banasandra komatiites

409 Jayananda et al. (2008) reported a Sm-Nd whole-rock isochron age of 3352 ± 110 410 Ma for komatiites collectively from various greenstone belts of western Dharwar craton, 411 including four samples from Banasandra. This age is based on the assumption that 412 komatiites exposed in different belts are all cogenetic starting with the same initial 413 143 Nd/ 144 Nd ratios.ACCEPTED We analyzed ten whole-rock samples from the Banasandra belt alone 414 which yielded an age of 3137 ± 190 Ma. The larger uncertainty associated with this age 415 makes it indistinguishable from the earlier reported age. However, when results of the 416 four samples of Banasandra from Jayananda et al. (2008) was combined with the present

15

ACCEPTED MANUSCRIPT

417 study, we get an age of 3139 ± 150 Ma (Fig. 14). Mukherjee et al. (2012) have also 418 obtained a 13 point whole-rock Sm-Nd isochron yielding an age of 3125 ± 120 Ma for 419 mafic-ultramafic suit of Nuggihalli greenstone belt which is considered to be a part of the 420 Sargur Group to which the Banasandra komatiites also belong to. It is, therefore, 421 suggested that Banasandra komatiites are of an age of ca. 3.14 Ga, and the komatiites in 422 other belts may have formed within couple of hundred million years of this time.

423 This age is also in agreement with the previous geochronological studies on the 424 supracrustals of Sargur greenstone belts. Nutman et al. (1992) using detrital zircons (U- 425 Pb SHRIMP) have derived an age range of 3.13–2.96 Ga for the formation of Sargur 426 supracrustals while Peucat et al. (1995) dated zircons from a rhyolite flow and proposed 427 an age of 3298 ± 7 Ma for Holenarsipur greenstone belt. Taken together, these ages 428 indicate protracted duration of development of Sargur greenstone belts, encompassing the 429 komatiitic flows.

430 5.3. Constraints from 146 Sm-142 Nd 431 The short lived isotopic system 146 Sm-MANUSCRIPT142 Nd has proved its potential, for 432 deciphering the details of early mantle differentia tion. (Regelous and Collerson, 1996; 433 Caro et al., 2003, 2006; Boyet and Carlson, 2005; Bennett et al., 2007; O’Neil et al., 434 2008; Brandon et al., 2009; Upadhyay et al., 2009; Caro, 2010; Murphy et al., 2010). 435 Recently Kinoshita et al. (2012) re-determined the half-life of 146 Sm and confirmed it to 436 be shorter (68 ±7 Ma) than the previous estimate of 103 Ma.

437 The main impetus to these studies came after the startling report that there is a 20 438 ppm deficit in 142 Nd/ 144 Nd ratios of chondrites relative to terrestrial standards (Boyet and 439 Carlson, 2005). Lu-Hf studies that followed the initial reports of positive 142 Nd reported 440 evidences of variable preservation of the early differentiation records due to probable 441 mixing of theACCEPTED early reservoirs with time (Bennett et al., 1993; Vervoort and Blichert-Toft, 442 1998). The boninites of Isua greenstone belts have also provided evidences of 443 differentiation of Hadean mantle (Frei et al., 2004).

16

ACCEPTED MANUSCRIPT

444 The results reporting the negative 142 Nd anomaly (O’Niel et al., 2008; Upadhyay 445 et al., 2009; Rizo et al., 2012) though raise several questions on the mantle convection, 446 preservation and mixing of chemical heterogeneities, do provide proofs of a 447 complementary enriched reservoir. This premise of early mantle-heterogeneity was tested 448 by Blichert-Toft and Puchtel (2010) by analyzing the Sm-Nd and Lu-Hf isotope results 449 from the various Archean komatiites. Their study yielded two interesting results: one 450 observation was that Archean mantle has been spatially homogenous as the komatiites of 451 the same age from different parts of the world yielded same time-integrated Sm/Nd 452 ratios. The second observation was that the depletion in the mantle has been decreasing 453 with time based on the fact that the time integrated Sm/Nd ratios decrease with the 454 decrease in age of the komatiites. It is in this context that our measurements of 455 142,143 Nd/ 144 Nd ratio in the Archean komatiites of Dharwar craton are of added 456 significance.

457 The samples we analysed possess 142 Nd/ 144 Nd values (Table 3) similar to that of 458 the terrestrial standards analysed with the samples (1.141874 ± 3) which shows that the 459 Banasandra komatiites formed from a source which had separated from the parental 146 MANUSCRIPT 460 reservoir after the extinction of Sm. The crystallization or emplacement age of this 461 komatiitic lava is ~3.14 Ga as yielded by the Sm-Nd whole-rock isochron of 10 samples 462 of the Banasandra belt. The isochron also provides constraints on nature of the source, by

463 means of the initial isotope ratio of +3.5 in εNd units. This positive value of εNd indicates 464 depletion of mantle which seems to be unexpectedly high compared to the other mantle 465 provinces during that time (Blichert-Toft and Puchtel, 2010). However, many other

466 studies have reported similar εNd or even higher values from the Dharwar craton in 467 Archean (Balakrishnan et al., 1990; Mukherjee et al., 2012). Dey (2013) compiled the Nd

468 isotope ratios of Dharwar craton and shown that the initial εNd values go as high as +6 in 469 Archean. However, these ratios alone do not provide constraints to quantify the depletion 470 events in termsACCEPTED of Sm/Nd fractionation. We have carried out 146,147 Sm-142,143 Nd combined 471 study to understand the mantle-differentiation better. The 147 Sm-143 Nd age of the samples 472 together with the absence of any radiogenic anomaly in 142 Nd/ 144 Nd ratio enables us to 473 calculate the 147 Sm/ 144 Nd ratio in the source during the period between 4.3 to 3.14 Ga. 17

ACCEPTED MANUSCRIPT

474 We have taken the beginning of this period at 4.3 Ga based on the fact that by this time 475 the 146 Sm would have been extinct, as per the new estimate of the half-life (Kinoshita et 476 al., 2012), causing no anomaly in the 142 Nd/ 144 Nd ratio. Thus the initial 143 Nd/ 144 Nd ratio

477 in the komatiite samples evolved to + 3.5 εNd during the period 4.3 to 3.14 Ga.

478 If the source is assumed to have separated from CHUR with present day 479 147 Sm/ 144 Nd = 0.1967 and 143 Nd/ 144 Nd= 0.512638 then the source would have had 480 147 Sm/ 144 Nd= 0.2222 and if the source is assumed to have separated from a previously 481 depleted mantle as per De Paolo’s depleted-mantle-model (De Paolo, 1981) with present 482 day 147 Sm/ 144 Nd= 0.2135 and 143 Nd/ 144 Nd= 0.513152 then the source would have had a 483 ratio of 147 Sm/ 144 Nd= 0.2178 (Fig. 15). In both cases the mantle depletion is higher than 484 even the present day mantle. This is the minimum estimate of the depletion as the age-

485 corrected εNd values of individual samples go as high as +6 units for some samples. This 486 differentiation occurred after the extinction of 146 Sm, i.e., at or later than 4.3 Ga. 487 However, the positive anomaly recorded by Isua greenstone rocks indicates mantle 488 differentiation at 4.53 Ga (Boyet and Carlson, 2005). This first differentiation left the ε 147 144 489 mantle with Nd = +2 and Sm/ Nd ratio of 0.2MANUSCRIPT from which the Isua greenstones would 490 have formed at ~3.8 Ga. The Banasandra komatiites at ca. 3.14 Ga with εNd = +3.5 are 491 sourced from mantle that has been differentiated a second time at 4.3 Ga (Fig. 16).

492 These two events would have generated complementary enriched reservoirs 493 which would have mixed variably to reduce the extent of depletion (Fig. 16). A model on 494 similar lines was proposed by Rizo et al. (2012) to explain the 142 Nd deficit in the 3.4 Ga 495 old Ameralik dykes in the Isua, Greenland which intrude the 3.8 Ga old Isua gneisses 142 496 with positive excess Nd. These dykes have positive initial εNd compared to the CHUR 497 but deficit in 142 Nd compared to the modern terrestrial samples. This apparent decoupling 498 of 142 Nd/ 144 Nd and 143 Nd/ 144 Nd is explained by invoking two events of mantle 499 differentiation,ACCEPTED first at ~4.47 Ga to give rise to the enriched source with negative 142 Nd 500 and the second mantle differentiation occurred ~770 Ma after the earth formation, to give 501 rise to the observed depletion in the 143 Nd/ 144 Nd (Rizo et al., 2012).

18

ACCEPTED MANUSCRIPT

502 The timing of the second differentiation was estimated keeping in view the 503 maximum Sm/Nd fractionation possible during melting to generate magmas represented 504 by Ameralik dykes. However, in the present case if we take 3.8 Ga to be the time of 505 second differentiation then the Sm/Nd ratio in mantle source for Banasandra komatiites 506 would be even higher at ~0.2385. Therefore, we propose that the second differentiation 507 would have occurred soon after the extinction of 146 Sm, i.e. at 4.3 Ga as evidenced by the 508 absence of 142 Nd anomaly.

509 In contrast to Blichert Toft and Puchtel (2010) model, it is observed that the time 510 integrated Sm/Nd ratio of Banasandra komatiites is much higher than that of komatiites 511 from any other region (Fig. 15) indicating spatial heterogeneity in the mantle during 512 Archean. This implies that some regions of mantle preserved these heterogeneities and 513 corroborates the interpretation made by Touboul et al. (2012) who found that there is 514 some excess in 182 W when compared with the modern terrestrial samples in the 2.8 Ga 515 old Kostomuksha komatiites. These heterogeneities should have been generated during 516 the differentiation events, which could be either global or local in nature. 517 6. Conclusions MANUSCRIPT 518 Komatiites with spinifex textures are exposed in the western Dharwar craton 519 within the Archean Sargur group of rocks. The samples studied during the present work 520 are parts of Banasandra greenstone belt which lies within Kibbanahalli arm of the 521 Chitradurga greenstone belt. Major and trace element abundance data brings out some 522 unique features of these komatiites which make them different from the AUK (e.g. 523 Munro type) or ADK (e.g. Barberton type). Particularly certain characteristic ratios such

524 as CaO/Al 2O3, Al 2O3/TiO 2, Gd/Yb and Nb/Y indicate them to be similar to AEK. The 525 wide range of the elemental and isotope ratios can be explained by the sequential melting 526 model. Starting with the primitive mantle composition, two episodes of partial melting 527 would generateACCEPTED a source which on subsequent melting would give rise to the observed 528 range of Sm/Nd ratios. Applying the conventional CMAS diagram yield a variable depth 529 ranging from 120–250 km. This pressure estimate, together with the trace-element

19

ACCEPTED MANUSCRIPT

530 modelling suggests that the source of these komatiites has already been depleted by melt- 531 extraction events.

532 Coupled chronometer of 146,147 Sm-142,143 Nd suggests that these komatiites 533 crystallized at 3.14 Ga from a source that was separated from the initial bulk silicate 534 reservoir at or later than ca. 4.3 Ga as evidenced by the absence of radiogenic 142 Nd 535 anomaly. These time constraints enabled the calculation of time-integrated 147 Sm/ 144 Nd 536 ratio which could be as high as 0.2222 if the parent silicate reservoir is CHUR or could 537 be 0.2178 if the parent reservoir is depleted mantle similar to that of De Paolo (1981). 538 Both these estimates are significantly higher compared to the estimates derived from the 539 other komatiite provinces of that period (Blichert-Toft and Puchtel, 2010) and point to 540 preservation of mantle heterogeneities caused by differentiation events.

541 This apparent discordance in the estimate of mantle-depletion can be reconciled 542 only if there is at least one more mantle-differentiation event younger than the initial 543 global differentiation at ca. 4.53 Ga (Boyet and Carlson, 2005). We here propose that the 544 two events of mantle differentiation not only explains the Nd isotope ratios of the 545 Banasandra komatiites, but, it also explains MANUSCRIPT the progressive decrease in the mantle 546 depletion with time due to mixing back of these enriched reservoirs generated during the 547 two events of differentiation. It, however, is difficult to ascertain whether the second 548 mantle-differentiation is global or local in nature.

549 Acknowledgements

550 RB thanks PLANEX (Planetary Exploration) program, Department of Space, India for 551 funding this study (PLANEX Ref. No. 5940). This manuscript is improved based on the 552 reviews and editorial suggestions on earlier versions of it and we thank Maud Boyet, 553 A.C. Kerr and three anonymous reviewers, for this.

554 ACCEPTED

20

ACCEPTED MANUSCRIPT

555 References

556 Anil Kumar, Bhaskar Rao, Y. J., Sivaraman, T. V., Gopalan, K., 1996. Sm–Nd ages of 557 Archaean metavolcanics of the Dharwar craton, South India. Precambrian Research 558 80, 205–216. 559 Arndt, N.T., 1975. Ultramfic rocks of Munro Township and their volcanic settings. Ph.D. 560 thesis. University of Toronto. 561 Arndt, N. T., Naldrett, A. J., Pyke, D. R., 1977. Komatiitic and iron rich tholeiitic lavas 562 of Munro Township, northeast Ontario. Journal of Petrology 18, 319-369. 563 Arndt, N.T., 1986. Differentiation of komatiitic flows. Journal of Petrology 27, 279-301. 564 Balakrishnan, S., Hanson, G.N., Rajamani, V., 1990. Pb and Nd isotope constraints on 565 the origin of high Mg and tholeiitic amphibolites, Kolar schist belt, south India. 566 Contributions to Mineralogy and Petrology 107, 279-292. 567 Balakrishnan, S., Rajamani, V., Hanson, G. N., 1999. U–Pb ages for zircon and titanite 568 from the Ramagiri area, southern India: Evidence for the accretionary origin of the 569 Eastern Dharwar craton during the late Archean. Journal of Geology 107(1), 69–86. 570 Barnes, S. J., Hill, R. E. T., Evans, N. J., 2004.MANUSCRIPT Komatiites and nickel sulfide ores of the 571 Black Swan area, Yilgarn Craton, Western Australia. 3: Komatiite geochemistry, and 572 implications for ore forming processes. Mineralium Deposita 39, 729- 751. 573 Beckinsale, R. D., Drury, S. A., Holt, R. W., 1980. 3360 Myr old gneisses from the south 574 Indian Craton. Nature 283, 469–470. 575 Bennett, V. C., Brandon, A. D., Nutman, A. P., 2007. Coupled 142 Nd–143 Nd evidence for 576 Hadean mantle dynamics. Science 318, 1907–1910. 577 Bennett, V. C., Nutman, A. P., McCulloch, M. T., 1993. Nd isotopic evidence for transient, 578 highly depleted mantle reservoirs in the early history of the earth. Earth and Planetary 579 Science Letters 119, 299- 317. 580 Blichert-Toft, J., Puchtel, I. S., 2010. Depleted mantle sources through time: Evidence from 581 Lu–Hf andACCEPTED Sm–Nd isotope systematics of Archean komatiites. Earth and Planetary 582 Science Letters 297, 598-606.

21

ACCEPTED MANUSCRIPT

583 Blicher-Toft, J., Arndt, T.N., Wilson, A., Coetzee, G., 2015. Hf and Nd isotope systematic 584 of early Archean komatiites from surface sampling and ICDP drilling in Barberton 585 Greenstone Belt, South Africa. American Mineralogist 100, 2396-2411. 586 Boyet, M., Carlson, R. W., 2005. 142 Nd evidence for early (>4.53 Ga) global differentiation 587 of the silicate Earth. Science 309, 576– 581. 588 Brandon, A. D., Lapen, T. J., Debaille, V., Beard, B. L., Rankenburg, K., Neal, C. R., 2009. 589 Re-evaluating 142 Nd/ 144 Nd in lunar mare basalts with implications for the early 590 evolution and bulk Sm/Nd of the Moon. Geochimica et Cosmochimica Acta 73, 6421- 591 6445. 592 Caro, G., 2010. Early silicate Earth differentiation. Annual Reviews on Earth and Planetary 593 Sciences 39, 31-58. 594 Caro, G., Bourdon, B., Birck, J. L., Moorbath, S., 2003. 146 Sm–142 Nd evidence from Isua 595 metamorphosed sediments for early differentiation. Nature 423, 428–432. 596 Caro, G., Bourdon, B., Birck, J. L., Moorbath, S., 2006. High precision 142 Nd/ 144 Nd 597 measurements in terrestrial rocks: constraints on the early differentiation of the Earth’s 598 mantle. Geochimica et Cosmochimica Acta 70, 164- 191. 599 Chadwick, B., Vasudev, V. N., Krishna Rao, MANUSCRIPT B., Hegde, G.V., 1992. The Dharwar 600 supergroup: Basin development and implications for Late Archean tectonic setting in 601 western India, southern India. In: Glover, J. E., Ho, S. (Eds.), The Archean: Terrains, 602 Processes and Metallogeny. University of Western Australia, Publ. 22, 3–15. 603 Chardon, D., Jayananda, M., Chetty, T. R. K., Peucat, J-J., 2008. Precambrian continental 604 strain and shear zone patterns: South Indian case. Journal of Geophysical Research 605 113 B08402, 1–16. 606 Chardon, D., Jayananda, M., Peucat, J.J., 2011. Lateral constrictional flow of hot 607 orogenic crust: insights from the Neoarchean of south India, geological and 608 geophysical implications for orogenic plateaux. Geochemistry, Geophysics, 609 GeosystemsACCEPTED 12, Q02005, http://dx.doi.org/10.1029/2010GC003398 610 Chavagnac, V., 2004. A geochemical and Nd isotopic study of Barberton komatiites 611 (South Africa): implication for the Archean mantle. Lithos 75, 253- 281.

22

ACCEPTED MANUSCRIPT

612 Cheng, S.H., Kusky, T.M., 2007. Komatiites from West Shardong, North China craton: 613 Implications for plume tectonics. Gondwana Research 12, 77-83. 614 Condie, K. C., 2003. Incompatible element ratios in oceanic basalts and komatiites: 615 tracking deep mantle source and continental growth rates with time. Geochemistry, 616 Geophysics, Geosystems 4(1), 1005. 617 Cox, K.G., Bell, J.D., Pankhurst, R.J., 1984. The Interpretation of Igneous Rocks. George 618 Allen and Unwin (Publishers) Ltd., London, UK. 619 De Paolo, D. J., 1981. Neodymium isotopes in the Colorado Front Range and crust-mantle 620 evolution in the Proterozoic. Nature 291, 193-197. 621 Dey, S., 2013. Evolution of Archaean crust in the Dharwar craton: The Nd isotope record. 622 Precambrian Research 227, 227- 246. 623 Donaldson, C.H., 1982. Spinifiex textured komatiites: A review of textures, composition 624 and layering. In: Arndt, N.T., Nisbet, E.G. (Eds.), Komatiites. Allen and Unwin 625 London, 213-244. 626 Frei, R., Polat, A., Meibom, A., 2004. The hadean upper mantle conundrum: Evidence for 627 source depletion and enrichment from Sm-Nd, Re-Os, and Pb isotopic composition in 628 3.71 Gy boninite-like metabasalts from MANUSCRIPT the Isua Supracrustal Belt, Greenland. 629 Geochimica et Cosmochimica Acta 68, 1645- 1660. 630 Hanski, E., Huhma, H., Rastas, P., Kamenetsky, V. S., 2001. The palaeoproterozoic 631 komatiite-picrate association of Finnish Lapland. Journal of Petrology 42 (5), 855-876. 632 Hokada, T., Horie, K., Satish Kumar, M., Ueno, Y., Nasheeth, A., Mishima, K., Shiraishi, 633 K., 2013. An appraisal of Archaean supracrustal sequences in Chitradurga Schist Belt, 634 Western Dharwar Craton, southern India. Precambrian Research 227, 99-119. 635 Hollings, P., Wyman, D., 1999. Trace element and Sm–Nd systematics of volcanic and 636 intrusive rocks from the 3 Ga Lumby Lake Greenstone belt, Superior Province: 637 evidence for Archean plume–arc interaction. Lithos 46, 189- 213. 638 Jaffri, S. H.,ACCEPTED Subba Rao, D.V., Ahmad, S. M., Mathur, R., 1997. Spinifex textured 639 peridotitic komatiites from Nuggihalli and Holenarsipur schist belts, Karnataka. 640 Journal of Geological Society of India 49, 33–38.

23

ACCEPTED MANUSCRIPT

641 Jayananda, M., Kano, T., Peucat, J.-J, Channabasappa, S., 2008. 3.35Ga komatiite 642 volcanism in the western Dharwar craton, southern India: Constraints from Nd isotopes 643 and whole rock geochemistry. Precambrian Research 162, 160-179. 644 Jensen, L.S., 1976. A new cation plot for classifying subalkaline volcanic rocks, Ontario 645 division of mines. Miscellaneous paper 66, 22. 646 Jochum, K. P., Arndt, N. T., Hofmann A. W., 1991. Nb-Th-La in komatiites and basalts: 647 constraints on komatiite petrogenesis and mantle evolution. Earth & Planetary 648 Science Letters 107, 272- 289. 649 Kinoshita, N., Paul, M., Kashiv, Y., Collon, P., Deibel, C. M., DiGiovine, B., Greene, J. 650 P., Henderson, D. J., Jiang, C. L., Marley, S.T., Nakanishi, T., Pardo, R. C., Rehm, K. 651 E., Robertson, D., Scott, R., Schmitt, C., Tang, X. D., Vondrasek, R., Yokoyama, A., 652 2012. A shorter 146 Sm half-life measured and implications for 146 Sm-142 Nd 653 chronology in the solar system. Science 335, 1614-1617. 654 Lecuyer, C., Gruau, G., Anhaeusser, C. R., Fourcade, S., 1994. The origin of fluids and 655 the effects of metamorphism on the primary chemical compositions of Barberton 656 komatiites: New evidence from geochemical (REE) and isotopic (Nd, O, H, 657 39 Ar/ 40 Ar) data. Geochimica et Cosmochimica MANUSCRIPT Acta 58, 969- 984. 658 Lahaye, Y., Arndt, N., Byerly, G., Chauvel, C., Fourcade, S., Gruau, G., 1995. The 659 influence of alteration on the trace-element and Nd isotopic compositions of 660 komatiites. Chemical Geology 126, 43-64. 661 Ludwig, K.R., 2001. User’s Manual for Isoplot/Ex rev. 2.49, A geochronological toolkit 662 for Microsoft Excel. Special Publication No. la, Berkeley Geochronology Center, 58. 663 Mazumder, R., Eriksson, P. G., 2015. Precambrian basins of India: Stratigraphic and 664 tectonic context. Geological Society London Memoir 43,185-205. 665 McDonough, W. F., Sun, S. S., 1995. The composition of the Earth. Chemical Geology 666 120, 223-253. 667 McKenzie, D.,ACCEPTED Bickle, M. J., 1988. The volume and composition of melt generated by 668 extension of the lithosphere. Journal of Petrology 29, 625-679. 669 Mukherjee, R., Mondal, S. K., Frei, R., Rosing, M. T., Waight, T. E., Zhong, H. Kumar, 670 G. R. R., 2012. The 3.1 Ga Nuggihalli chromite deposits, Western Dharwar craton

24

ACCEPTED MANUSCRIPT

671 (India): Geochemical and isotopic constraints on mantle sources, crustal evolution 672 and implications for supercontinent formation and ore mineralization. Lithos 155, 673 392- 409. 674 Murphy, D. T., Brandon, A. D., Debaille, V., Burgess, R., Ballentine, C., 2010. In search 675 of a hidden long term isolated sub-chondritic 142 Nd/ 144 Nd reservoir in the deep 676 mantle: Implications for the Nd isotope systematics of the Earth. Geochimica et 677 Cosmochimica Acta 74, 738-750. 678 Nesbitt, R.W., Sun, S-S, Purvis, A.C.,1979. Komatiites: Geochemistry and Genesis. 679 Canadian Mineralogist 17, 165-186. 680 Nutman, A. P., Chadwick, B., Ramakrishnan, M., Viswanatha, M. N., 1992. SHRIMP U– 681 Pb ages of detrital zircon in Sargur supracustal rocks in western Karnataka, Southern 682 India. Journal of Geological Society of India 39, 367–374. 683 Nutman, A. P., Chadwick, B., Krishna Rao, B., Vasudev, V. N., 1996. SHRIMP U–Pb 684 zircon ages of acid volcanic rocks in the Chitradurga and Sandur Groups and 685 granites adjacent to Sandur schist belt. Journal of Geological Soceity of India 47, 686 153–161. 687 O’Hara, M. J., 1968. The bearing of phase equilibriMANUSCRIPTa studies in synthetic and natural 688 systems on the origin and evolution of basic and ultrabasic rocks. Earth Science 689 Reviews 4, 39-133. 690 O’Neil, J., Carlson, R.W., Don Francis, Ross K. Stevenson, 2008. Neodymium-142 691 evidence for Hadean mafic crust. Science 321, 1828–1831. 692 Ohtani, E., 1990. Majorite fractionation and genesis of komatiites in the deep mantle. 693 Precambrian Research 48, 195- 202. 694 Peucat, J-J, Bouhallier, H, Fanning, C. M., Jayananda, M., 1995. Age of the 695 Holenarsipur greenstone belt, relationships with the surrounding gneisses 696 (Karnataka, South India). Journal of Geology 103, 701–710. 697 Peucat J. J, ACCEPTED Mahabaleshwar B. and Jayananda M., 1993. Age of younger tonalitic 698 magmatism and granulitic metamorphism in the South India transition zone 699 (Krishnagiri area): Comparison with older Peninsular gneisses from the Gorur- 700 Hassan area. Journal of Metamorphic Geology 11, 879–888.

25

ACCEPTED MANUSCRIPT

701 Polat, A., Kerrich, R., Wyman, D. A., 1999. Geochemical diversity in oceanic komatiites 702 and basalts from the late Archean Wawa greenstone belts, Superior Province, 703 Canada: trace element and Nd isotope evidence for a heterogeneous mantle. 704 Precambrian Research 94, 139- 173. 705 Geological Survey of India Publication, 1994. Project Vasundhara Map. 706 Puchtel, I. S., Humayun, M., 2005. Highly siderophile element geochemistry of 187 Os- 707 enriched 2.8 Ga Kostomuksha komatiites, Baltic Shield. Geochimica et Cosmochimica 708 Acta 69 (6), 1607- 1618. 709 Radhakrishna, B. P., Sreenivasaiah, C., 1974. Bedded barites from the Precambrian of 710 Karnataka. Journal of Geological Soceity of India 15, 314-315. 711 Rajamani, V., Shivkumar, K., Hanson, G. N., Shirey, S. B., 1985. Geochemistry and 712 petrogenesis of amphibolites, Kolar Schist Belt, south India: Evidence for komatiitic 713 magma derived by low percentages of melting of the mantle. Journal of Petrology 714 26(1), 92-123. 715 Regelous, M., Collerson, K. D., 1996. 147 Sm-143 Nd, 146 Sm-142 Nd systematics of Early 716 Archean rocks and implications for crust-mantle evolution. Geochimica et 717 Cosmochimica Acta 60, 3513-3520. MANUSCRIPT 718 Revillon, S., Arndt, N. T., Chauvel, C., Hallot, E., 2000. Geochemical study of ultramafic 719 volcanic and plutonic rocks from Gorgona Island, Colombia: the plumbing system of an 720 oceanic plateau. Journal of Petrology 41 (7), 1127-1153. 721 Rizo, H., Boyet, M., Blichert-Toft, J., O’Neil, J., Rosing, M. T., Paquette, J-L., 2012. The 722 elusive Hadean enriched reservoir revealed by 142 Nd deficits in Isua Archaean rocks. 723 Nature 491, 96-100. 724 Robin-Popieul, C.C.M., Arndt, N.T., Pik, R., Martineau, F., Fourcade, S., Marty, B. 725 2012. A new model for Barberton komatiites: deep critical melting with high melt 726 retention. Journal of Petrology 53, 2191-2229. 727 Seshadri, T. ACCEPTEDS., Chaudhuri, A., Harinadha Babu, P., Chayapathi, N., 1981. Chitradurga 728 belt. In: J. Swaminath, M. Ramakrishnan (Eds.), Early Precambrian Supracrustals of 729 Southern Karnataka. Geological Survey of India Memoirs 112, 163–198.

26

ACCEPTED MANUSCRIPT

730 Shapiro, L., Brannock, W. W., 1962. Rapid analyses of silicate, carbonate and phosphate 731 rocks. Bulletin of U. S. Geological Survey 1144A. 732 Sheth, H. C., Ray, J. S., Bhutani, R., Kumar , A., Smitha R. S., 2009. Volcanology and 733 eruptive styles of Barren Island: an active mafic strato volcano in the Andaman Sea, NE 734 Indian Ocean. Bulletin of Volcanology, Doi: 10.1007/s00445-009-0280-z. 735 Sproule, R. A., Lesher, C. M., Ayer, J. A., Thurston, P. C., Herzberg, C. T., 2002. Spatial 736 and temporal variations in the geochemistry of komatiites and komatiitic basalts in the 737 Abitibi greenstone belt. Precambrian Research 115, 153- 186. 738 Srikantia, S. V., Bose, S. S., 1985. Archaen komatiites from Banasandra area of 739 Kibbanahalli arm of Chitradurga Supracrustal belt in Karnataka. Journal of Geological 740 Soceity of India 26, 407-417. 741 Sun, S-s, McDonough, W.F. 1989. Chemical and isotopic systematics of oceanic basalts: 742 implications for mantle composition and processes. Geological Society, London, 743 Special Publications 42, 313-345. 744 Takahashi, E., 1986. Melting of dry peridotite KLB1 up to 14 GPa: implications on the 745 origin of perioditie upper mantle. Journal of Geophysical Research 91, 9367-9382. 746 Takahashi, E., Scarfe, C. M., 1985. Melting ofMANUSCRIPT peridotite to 14 GPa and the genesis of 747 komatiite. Nature 315, 566-568. 748 Touboul, M., Puchtel, I. S., Walker, R. J., 2012. 182 W evidence for long-term preservation 749 of early mantle differentiation products. Science 335, 1065-1069. 750 Trendall, A. F., de Laeter, J. R., Nelson, D. R., Bhaskar Rao, Y. J., 1997. Further zircon 751 U–Pb age of the Daginkatte Formation, Dharwar Supergroup, Karnataka craton. 752 Journal of Geological Society of India 50, 25–30. 753 Tushipokla, Jayananda, M., 2013. Geochemical constraints on komatiite volcanism from 754 Sargur Group Nagamangala greenstone belt, western Dharwar craton, southern India: 755 Implications for Mesoarchean mantle evolution and continental growth. Geoscience 756 Frontiers ACCEPTED4(3), 321-340. Doi: 10.1016/j.gsf.2012.11.003. 757 Upadhyay, D., Scherer, E. E., Mezger, K., 2009. 142 Nd evidence for an enriched Hadean 758 reservoir in cratonic roots. Nature 459, 1118–1121.

27

ACCEPTED MANUSCRIPT

759 Van Acken, D., Hoffmann, J.E., Schorsher, J.H.D, Schulz, T., Heuser, A., Luhuet, A., 760 2016. Formation of high Al-komatiites from the Mesoarchean Quebra Osso Group, 761 Minas Gerais, Brazil: Trace elements, HSE systematics and Os isotope signatures. 762 Chemical Geology 422, 108-121. 763 Vervoort, J. D., Blichert-Toft, J., 1998. Evolution of the depleted mantle: Hf isotope 764 evidence from juvenile rocks through time. Geochimica et Cosmochimica Acta 63 , 765 533- 556. 766 Viljoen M. J. and Viljoen R. P., 1969a. The geology and geochemistry of the lower 767 ultramafic unit of the Onverwacht Group and a proposed new class of igneous 768 rock. Geol. Soc. S.Africa Spec. Publ . No.2, 85-86. 769 Viljoen M. J. and Viljoen R. P., 1969b. Evidence for the existence of a mobile extrusive 770 peridotitic magma from Komati Formation of Onverwacht Group. Geol. Soc. S.Africa, 771 Spec. Publ. No. 2, 87-112. 772 Viswanatha, M. N., Ramakrishnan, M., Narayanan Kutty, T. R., 1977. Possible spinifex 773 texture in a serpentinite from Karnataka. Journal of Geological Society of India 18, 774 194–197. 775 Viswanathan, S., 1974. Contemporary trends inMANUSCRIPT geochemical studies of early Precambrian 776 granite-greenstone complexes. Journal of Geological Society of India 15, 347- 379. 777 Xie, Q., McCuaig, T. C., Kerrich, R., 1995. Secular trends in the melting depths of mantle 778 plumes: evidence from HFSE/REE systematics of Archean high-Mg lavas and modem 779 oceanic basalts. Chemical Geology 126, 29- 42. 780 781 782 783 784 785 ACCEPTED 786 787 788

28

ACCEPTED MANUSCRIPT

789 790 Figure Captions: 791 Figure 1. Geological map of Dharwar craton showing Chitradurga Mylonitic Zone 792 (CMZ) that separates the craton into western and eastern parts. Greenstone belts of 793 Sargur Group are shown in the figure: 1. Ghattihosahalli, 2. Jayachamarajapura (J. C. 794 Pura), 3. Banasandra, 4. Nuggihalli, 5. Kalyadi, 6. Holenarsipur. The study area, 795 Banasandra, is marked by the box. Map is modified after Vasundhara project of 796 Geological Survey of India Publication (1994). 797 798 Figure 2. Geological map of Banasandra area which shows main exposures of komatiite 799 near Birasandra, Kodihalli and Kunikenahalli villages. The sample locations are also 800 indicated (modified after Srikantia and Bose, 1985). 801 802 Figure 3. (a) Spinifex textured komatiite – long olivine blades are visible in an outcrop 803 near Birasandra; (b) Pillowed serpentinite bounded by cherty layers near Kunikenahalli; 804 (c) Photomicrograph (sample B6/1) in crossed polars, showing long olivine blades in 805 serpentine matrix. Recrystallisation of olivine MANUSCRIPT into actinolite is visible along the grain 806 boundary, width of the view is 0.53 mm; (d) Photomicrograph of thin-section of 807 komatiite sample B2/4 in crossed polars, showing the characteristic assemblage of 808 actinolite-serpentine-chlorite-epidote. Width of the photo is 2.1 mm; (e) Parallel polars 809 view of two basal sections of actinolite found in a matrix of chlorite and serpentine in 810 sample B2/4. Width of the photo is 2.1 mm; (f) Euhedral olivine crystals surrounded by 811 serpentine matrix in plane polarized light (sample B5/1). Width of the photo is 1.05 mm. 812 t 813 Figure 4. Concentrations of major element oxides–TiO 2, Al 2O3, CaO and Fe 2O3 in 814 komatiites from Banasandra plotted against MgO showing significant trends. Data from 815 Arndt et al.ACCEPTED (1977) (Munro); Jochum et al. (1991) (Onverwacht, Kambalda and 816 Belingwe); Lecuyer et al. (1994) (Schapenburg); Xie et al. (1995) (Boston); Hollings and 817 Wyman (1999) (Lumby lake, Superior Province); Polat et al. (1999) (Wawa); Revillon et 818 al. (2000) (Gorgona); Hanski et al. (2001) (Finnish Lapland); Sproule et al. (2002)

29

ACCEPTED MANUSCRIPT

819 (Abitibi); Barnes et al. (2004) (Yilgarn); Chavagnac V. (2004) (Barberton); Puchtel and 820 Humayun (2005) (Kostomuksha); Jayananda et al. (2008) (Banasandra*). 821 822 Figure 5. (a) Jensen’s cation plot, constructed using cation proportions calculated from 823 mole proportions of the samples, is used for the classification of komatiites from other 824 ultramafic rocks. Most of the samples from Banasandra plot in the field of komatiites 825 except two of them which are altered komatiitic basalts and another two samples are 826 tholeiites (after Jensen, 1976). (b) CMAS diagram for komatiites and tholeiites from 827 Banasandra indicates that the melt for these komatiites is derived from different depths 828 ranging from 3 to 6 GPa for the komatiites and near 8 GPa for the peridotitic komatiites 829 (after O’Hara, 1968). Two tholeiites plot between 0 and 3 GPa. 830 831 Figure 6. Chondrite normalised REE patterns of (a) komatiites and (b) tholeiites from 832 Banasandra. Patterns of the samples B1/2 and B4/2A indicate alteration of these samples. 833 834 Figure 7. Sm-Nd whole-rock isochron for komatiites from Banasandra greenstone belt. 835 One sample (B6/1), shown as ‘o’ is not included MANUSCRIPT in the isochron. 836 837 Figure 8. 142 Nd/ 144 Nd plot for Banasandra komatites and tholeiites (B8/1, B14/1) along 838 with komatiitic amphibolites from Kolar greenstone belt and a recent volcanic sample 839 from Barren Island. The error bars are 2 σ. The errors on the measurements of the isotope 840 standards, AMES and La Jolla are 2 σ external-errors on the mean. 841 842 Figure 9. Nb/Y vs. La/Nb plot of Banasandra komatiites plotted along with Aluminium 843 Depleted (ADK) and Aluminium Undepleted Komatiites (AUK) (Condie, 2003). All the 844 samples show consistent results. 845 ACCEPTED 846 Figure 10. (Gd/Yb) N ratio vs. Al 2O3/TiO 2 plot of komatiites from Banasandra along with 847 that of komatiites from other regions of world. 848

30

ACCEPTED MANUSCRIPT

849 Figure 11. Ni vs. Zr and Ni vs. Nd plots of Banasandra komatiites. The curve shows the 850 concentrations of these elements for different degrees of partial melting of a peridotitic 851 mantle leaving olivine-only residue. Most of the samples indicate high degree of partial 852 melting with two peridotitic-komatiites plotting near 100% melting indicating them to be 853 akin to the parental melt of these komatiites. It is inferred that magmas representing 854 komatiites were derived from mantle sources that had earlier undergone multiple events 855 of melting and melt extractions. 856 857 Figure 12. Sm vs. Nd plot showing results of partial melting of a residue which has 858 undergone two prior events of melting The inset shows evolution of source on extraction 859 of melt in two steps of 10% melting each, starting with a concentration of Sm= 0.444, 860 Nd=1.354 (Sun and McDonough, 1989). 861 862 Figure 13. A cartoon diagram depicting the generation and emplacement of Banasandra 863 komatiite by re-melting of a rising mantle starting at a depth of ~250 km. The melt 864 extracted during different episodes, now make a composite Banasandra komatiite belt. 865 The consequent change in the Sm/Nd ratio ofMANUSCRIPT residue, and Gd/Yb versus Al 2O3/TiO 2 866 variation in the melt is also shown corresponding to different stages of melting. 867 Figure 14. Sm-Nd whole-rock isochron diagram for ten samples of komatiite of the 868 present study along with the four samples of komatiite of Banasandra analysed by 869 Jayananda et al. (2008). This fourteen point isochron yields an age of 3139 ± 150 Ma 870 which is same but slightly more precise than the ten points isochron (Fig. 7). 871 872 Figure 15. Time integrated 147 Sm/ 144 Nd ratios of Banasandra komatiites along with the 873 ratios of the komatiites from other provinces around the world (Blichert-Toft and Puchtel 874 2010). Note that 147 Sm/ 144 Nd ratios of Banasandra komatiites are higher compared to 875 other komatiitesACCEPTED of similar age. 876 877 Figure 16. A schematic diagram illustrating the sequence of early-mantle differentiation, 878 consistent with the observed 142 Nd/ 144 Nd, and time-integrated 147 Sm/ 144 Nd ratios. The

31

ACCEPTED MANUSCRIPT

879 higher positive anomaly in the Isua greenstone belt is because of differentiation of the 880 source mantle before the extinction of the 146 Sm, while absence of it in the source of 881 Banasandra is due to the differentiation after the extinction of the 146 Sm, i.e. ~4.3 Ga. 882 883 884 885 886 887 888 Table Captions:

889 Table 1 Major and trace element data of komatiites and tholeiites from Banasandra area 890 of Sargur group, Western Dharwar Craton. Major elements are in wt.% and trace 891 elements in ppm.

892 Table 2 147 Sm-143 Nd isotope data on komatiites and tholeiites of Banasandra greenstone

893 belt, Dharwar craton. The εNd (t) values were calculated for t = 3140 Ma.

142 144 MANUSCRIPT 894 Table 3 Nd/ Nd isotope composition of komatiites and tholeiites from Banasandra 895 greenstone belt and komatiitic amphibolites from Kolar greenstone belt, Dharwar craton 896 along with that of a sample from Barren Island, Andaman. Results from Nd isotope- 897 standards AMES and La Jolla analyzed with the samples are given for comparison.

ACCEPTED

32

ACCEPTED MANUSCRIPT

Rock type Komatiites Tholeiites Sample B 1/2 B1/3 B 2/3 B 2/4 B 3/1 B 4/2A B 5/1 B 6/1 B 10/1 B 10/2 B11/1 B 8/1 B 14/1

SiO2 28.74 43.04 44.47 38.75 38.04 30.03 37.18 44.76 41.9 43.76 37.61 52.05 51.77

TiO2 0.35 0.11 0.2 0.06 0.35 0.42 0.1 0.17 0.19 0.12 0.14 1.06 0.71

Al2O3 17.42 7.02 10.64 10.03 8.99 17.39 2.98 6.22 6.76 4.7 3.5 13.47 13.68

Fe2O3 12.43 6.9 9.56 7.31 8.2 11.49 6.07 5.59 7.57 7.4 8.28 9.72 8.94 MnO 0.1 0.1 0.15 0.17 0.1 0.11 0.11 0.09 0.13 0.11 0.11 0.16 0.13 MgO 26.16 27.37 24.64 26.61 24.92 27.54 34.06 25.89 24.46 30.42 34.07 5.19 5.15 CaO 0.1 3.36 4.73 4.7 3.94 0.07 2.37 5.74 5.62 2.34 0.98 7.6 8.51

K2O 0.11 0.02 0.04 0.06 0.12 0.03 0.04 0.06 0.06 0.04 0.04 0.32 0.22

Na2O 0.39 0.29 0.57 0.28 0.55 0.49 0.2 0.11 0.16 < 0.1 < 0.1 2.98 2.57

P2O5 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 0.01 LOI 13.34 13.14 7.14 10.19 12.69 14.19 17.58 12.27 10.25 12.6 14.41 6.62 6.1 Total 99.14 101.4 102.2 98.17 97.89 101.76 100.7 100.9 97.1 101.5 99.13 99.2 97.79 Mg# 0.81 0.89 0.84 0.88 0.86 0.82 0.92 0.9 0.86 0.89 0.89 0.51 0.53 Ba 31.31 6.35 10.2 15.85 30.17 3.66 MANUSCRIPT2.32 12.33 8.46 1.45 1 79.62 29.58 Cr 1429 3024 2959 2909 3654 6226 2557 3689 3942 1938 2206 188 159 Sc 63 24 41 37 35 39 13 22 28 21 14 31 35 V 295 108 174 154 177 257 69 73 138 106 96 270 230 Co 112 90 98 92 81 95 105 83 86 91 134 50 49 Ni 366 1216 732 1055 948 1254 2040 1147 1085 1657 2180 115 136 Zn 53 46 49 43 38 78 30 34 41 41 75 98 76 Rb 0.41 2.87 0.86 0.68 1.82 0.16 3.56 0.55 0.55 0.42 1.42 25.41 12.67 Sr 3.94 32.01 9.08 7.36 20 1.71 35.01 6.08 5.98 4.61 15.35 247 135 Y 1.09 2.98 6.03 10.01 5.02 1.82 3.39 2.46 4.44 2.77 5.02 45.06 21.59 Zr 10.99 0.48 3.58 1.77 6.58 22.15 0.7 2.2 1.7 1.07 0.64 73.34 37.45 Nb 0.81 0.24 0.47 0.09 0.48 3.76 0.26 0.21 0.36 0.25 0.4 9.68 5.72 La 1.18 0.21 0.25 0.12ACCEPTED0.49 1.6 0.34 0.23 0.31 0.19 0.64 23.82 9 Ce 3.1 0.52 0.68 0.01 0.76 3.5 0.7 0.46 0.44 0.27 1.51 49.57 17.36 Pr 0.5 0.11 0.17 0.04 0.26 0.46 0.13 0.09 0.12 0.08 0.24 6.92 2.66 Nd 2.27 0.56 0.94 0.23 1.44 1.9 0.61 0.44 0.6 0.43 1.21 27.4 10.32 ACCEPTED MANUSCRIPT

Sm 0.45 0.24 0.39 0.13 0.56 0.39 0.24 0.18 0.26 0.21 0.41 6.58 2.53 Eu 0.03 0.07 0.12 0.08 0.17 0.05 0.11 0.06 0.09 0.08 0.15 1.78 0.77 Gd 0.35 0.26 0.54 0.33 0.7 0.44 0.33 0.27 0.41 0.27 0.57 7.57 3.11 Tb 0.04 0.06 0.12 0.12 0.15 0.07 0.07 0.06 0.1 0.07 0.12 1.24 0.53 Dy 0.18 0.49 1.04 1.35 1.09 0.36 0.57 0.45 0.84 0.55 0.86 8.12 3.75 Ho 0.04 0.1 0.23 0.35 0.2 0.07 0.12 0.1 0.18 0.12 0.17 1.62 0.78 Er 0.15 0.33 0.72 1.14 0.53 0.21 0.34 0.27 0.51 0.33 0.51 4.57 2.28 Tm 0.03 0.05 0.12 0.21 0.07 0.04 0.05 0.04 0.09 0.06 0.08 0.67 0.33 Yb 0.2 0.4 0.95 1.39 0.44 0.24 0.37 0.25 0.6 0.4 0.52 4.3 2.23 Lu 0.04 0.06 0.14 0.2 0.05 0.05 0.05 0.03 0.08 0.05 0.07 0.54 0.29 Hf 0.32 0.01 0.15 0.08 0.26 0.58 0.04 0.12 0.1 0.08 0.04 1.85 0.96 Pb 0.73 0.17 0.82 0.42 0.44 0.29 0.67 0.14 0.2 0.38 0.84 6.55 4.87 Th 0.22 0.05 0.11 0.05 0.07 0.22 0.06 0.05 0.12 0.08 0.06 3.29 2.43 U 0.02 0.01 0.04 0.01 0.01 0.07 0.02 0.01 0.01 0.01 0.02 0.92 0.5

(Al2O3/TiO2) 49.8 63.8 53.2 167.17 25.7 41.4 29.8 36.6 35.6 39.2 25 12.71 19.27

(CaO/Al2O3) 0.01 0.48 0.44 0.5 0.44 <0.01 0.8 0.92 0.83 0.5 0.28 0.56 0.62 (Gd/Yb)N 1.44 0.52 0.46 0.19 1.29 1.51MANUSCRIPT 0.72 0.87 0.55 0.55 0.89 1.42 1.13 Nb/Th 3.63 5.24 4.25 1.89 6.92 16.85 4.05 4.23 3.01 3.26 7.11 2.94 2.35 Zr/Nb 13.59 1.97 7.57 19.38 13.7 5.88 2.71 10.68 4.67 4.26 1.61 7.58 6.55 Zr/Y 10.11 0.16 0.59 0.18 1.31 12.17 0.21 0.89 0.38 0.39 0.13 1.63 1.73 Nb/Y 0.74 0.08 0.08 0.01 0.1 2.07 0.08 0.08 0.08 0.09 0.08 0.21 0.26

ACCEPTED 147 ACCEPTED144 143 MANUSCRIPT144 σ Sample Sm (ppm) Nd (ppm) Sm/ Nd Nd/ Nd 2 εNd(t ) fSm/Nd B1/2 0.1476 0.6467 0.138 0.511726 0.000025 6 -0.3 B1/3 0.1817 0.5125 0.2145 0.513052 0.000004 0.9 0.09 B2/3 0.3609 0.8697 0.251 0.513936 0.000003 3.4 0.28 B2/4 0.0989 0.153 0.3909 0.516755 0.000003 1.7 0.99 B3/1 0.4805 1.2393 0.2344 0.513697 0.000003 5.4 0.19 B4/2A 0.3582 1.7683 0.1224 0.511174 0.000002 1.5 -0.38 B5/1 0.2126 0.5759 0.2232 0.513223 0.000003 0.7 0.13 B6/1 0.1677 0.4158 0.2577 0.514673 0.000004 15.1 0.31 B10/1 0.2337 0.5284 0.2674 0.514376 0.000004 5.3 0.36 B10/2 0.2363 0.5349 0.2673 0.514395 0.000003 5.8 0.36 B11/1 0.3681 1.0677 0.2085 0.512945 0.000003 1.2 0.06 B8/1 5.8313 24.3818 0.1446 0.511652 0.000003 1.9 -0.27 B14/1 2.1939 9.065 0.1463 0.511633 0.000002 0.8 -0.26

MANUSCRIPT

ACCEPTED Sample Name Rock type Location ACCEPTED142Nd/144Nd MANUSCRIPT142Nd/144Nd-Error (1σ ) Standard 142Nd/144Nd 142Nd/144Nd-Error (1σ ) B1/2 Komatiite Banasandra 1.141868 0.000003 LaJolla-1 1.141878 0.000004 B1/3 Komatiite Banasandra 1.14187 0.000003 AMES-1 1.141871 0.000005 B2/3 Komatiite Banasandra 1.141876 0.000002 LaJolla-1 1.141874 0.000005 B3/1 Komatiite Banasandra 1.141873 0.000002 AMES-2 1.141877 0.000003 B3/1R Komatiite Banasandra 1.141876 0.000002 LaJolla-2 1.141878 0.000003 B4/2A Komatiite Banasandra 1.141875 0.000005 AMES-1 1.141874 0.000002 B5/1 Komatiite Banasandra 1.141869 0.000003 LaJolla-1 1.141879 0.000007 B5/1R Komatiite Banasandra 1.141872 0.000002 LaJolla-2 1.141878 0.000005 B6/1 Komatiite Banasandra 1.14187 0.000004 AMES-1 1.141876 0.000009 B10/1 Komatiite Banasandra 1.14187 0.000007 AMES-2 1.141872 0.000004 B10/1R Komatiite Banasandra 1.141867 0.000003 LaJolla-2 1.141869 0.000004 B11/1 Komatiite Banasandra 1.141879 0.000007 AMES-1 1.141876 0.000003 B8/1 Tholeiite Banasandra 1.141871 0.000004 AMES-2 1.141875 0.000004 B14/1 Tholeiite Banasandra 1.141874 0.000003 LaJolla-1 1.141871 0.000002 KGF-18-15 Komatiitic amphibolite Kolar 1.141867 0.000002 AMES-1 1.141873 0.000003 KGF_18-15R Komatiitic amphibolite Kolar 1.141869 0.000002 AMES-2 1.141876 0.000006 SB3-23 Komatiitic amphibolite Kolar 1.141871 0.000002 LaJolla-1 1.141868 0.000003 SB3-23R Komatiitic amphibolite Kolar 1.141863 0.000004 LaJolla-2 1.141871 0.000006 SB3-26 Komatiitic amphibolite Kolar 1.141868 0.000002 AMES-2 1.141868 0.000002 SB3-26R Komatiitic amphibolite Kolar 1.141874 0.000003 AMES-2 1.141868 0.000002 BI-0709 Basalt Barren Island 1.141873 0.000003 AMES-1 1.141872 0.000003 Average 1.141871 0.000004 Average 1.141874 0.000004

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

Highlights:

• Al-undepleted to enriched komatiite are generated by melting of residue at 3.15 Ga.

• Time-integrated Sm/Nd ratio of the source mantle higher than contemporary mantle.

• More than one differentiation events separated by space and time in Hadean mantle.

• Mantle heterogeneity generated in Hadean was preserved till Archean.

MANUSCRIPT

ACCEPTED