Heterogeneities from the first 100 million years recorded in deep mantle noble gases from the Northern Lau Back-arc Basin

Heavy noble gases (Ne, Ar, Xe) can record long-lasting heterogeneities in the mantle 11 because of the production of isotopes from extant ( 238 U, 40 K) and extinct ( 129 I and 244 Pu) 12 radionuclides. However, the presence of ubiquitous atmospheric contamination, particularly for 13 ocean island basalts (OIBs) that sample the Earth’s deep mantle, have largely hampered precise characterization of the mantle source compositions. Here we present new high-precision noble 15 gas data from gas-rich basalts erupted along the Rochambeau Rift in the northwestern corner of 16 the Lau Basin. The strong influence of a deep mantle plume in the Rochambeau source is 17 apparent from low 4 He/ 3 He ratios down to 25,600 ( 3 He/ 4 He of 28.1 R A ). We find that the 18 Rochambeau source is characterized by low ratios of radiogenic to non-radiogenic nuclides of 19 Ne, Ar, and Xe (i.e., low 21 Ne/ 22 Ne, 40 Ar/ 36 Ar, and 129 Xe/ 130 Xe) compared to the mantle source 20 of mid-ocean ridge basalts (MORBs). Additionally, we observe differences in elemental 21 abundance patterns between the Rochambeau source and the mantle source of MORBs as 22 characterized by the gas-rich popping rock from the Mid-Atlantic Ridge. However, the 3 He/ 22 Ne 23 ratio of the Rochambeau plume source is significantly higher than the Iceland and Galapagos 24 plume sources, while the 3 He/ 36 Ar and 3 He/ 130 Xe ratios appear to be similar. The difference in 25 3 He/ 22 Ne between Rochambeau and the Galapagos and Iceland plume sources could reflect long 26 lasting accretional heterogeneities in the deep mantle or some characteristic of the back-arc 27 mantle source. isotopic measurements indicate that the lower 129 Xe/ 130 Xe ratios in the Rochambeau source cannot be explained solely by mixing atmospheric xenon with type xenon; nor be to produce the compositions seen in higher proportion of Pu-derived fission Xe in the Rochambeau source compared to the MORB source. Therefore, both I/Xe and Pu/Xe ratios are different between OIB and MORB mantle sources. Our observations require heterogeneous volatile accretion and a lower degree of 35 processing for the mantle plume source compared to the MORB source. Since differences in 36 129 Xe/ 130 Xe ratios have to be produced while 129 I is still alive, OIB and MORB sources were 37 degassed at different rates for the first 100 Ma of Solar System history, and subsequent to this 38 period, the two reservoirs have not been homogenized. In combination with recent results from 39 the Iceland plume, our observations require the preservation of less-degassed, early-formed heterogeneities in the Earth’s deep mantle throughout Earth’s history.


43
The noble gas compositions of mantle-derived basalts provide information on the 44 degassing history, style of mantle convection, and volatile exchange between the deep Earth and 45 exosphere. Compared to mid-ocean ridge basalts (MORBs), ocean island basalts (OIBs) from 46 Iceland, Hawaii, Galapagos, Réunion and Samoa are characterized by lower ratios of radiogenic 47 to primordial isotopes such as 4 He/ 3 He, 21 Kunz, 2005). 61 If the low 129 Xe/ 130 Xe ratios in OIBs are indeed from a less degassed reservoir, then the 62 OIB and MORB reservoirs must be partially isolated from each other since 4.45 Ga as 129 I, 63 which produces 129 Xe, became extinct 100 million years after the start of the Solar System. Such 64 long-term separation would invalidate many models put forth to explain the chemical and 65 dynamical evolution of the mantle. On the other hand, if the differences in 129 Xe/ 130 Xe ratios in 66 OIBs are from recycling of atmospheric Xe, long-term separation of the two sources is not 67 required and extensive mixing between the sources is allowed. Hence, addressing the origin of  25,600 (28.1 R A , where R A is the 3 He/ 4 He ratio normalized to the atmospheric ratio of 1.39 × 10 -79 6 ). 80 The Rochambeau Rift is located in the northwestern flank of the Lau back-arc basin, 81 behind the Tonga arc, in the western Pacific (Fig. 1). Shear-wave splitting analyses suggest a fast 82 direction of anisotropy that is oriented north to south in the Lau back-arc spreading center (Smith Tonga slab beneath the Vitiaz lineament (Millen and Hamburger, 1998 range from 3 to 23 × 10 -6 cm 3 STP g -1 . Glass chunks were carefully selected to avoid 113 phenocrysts. In order to remove surface alteration, glasses were leached in 2% nitric acid for 10 114 to 20 minutes, and then ultrasonically cleaned in distilled water and acetone. Single pieces of 115 basaltic glass (3.2 to 6.8 grams) were baked under vacuum for 24 hours at 100ºC and were 116 pumped for an additional 6 to 12 days. Samples were crushed under vacuum using a hydraulic 117 ram to release magmatic gases trapped in vesicles. The released gases were purified by 118 sequential exposure to hot and cold SAES getters and a small split of the gas was let into a 119 quadrupole mass spectrometer to determine the Ar abundance and an approximate 40 Ar/ 36 Ar 120 ratio. The noble gases were then trapped on a cryogenic cold-finger. He was separated from Ne 121 at 32 K and let into the Nu Noblesse mass spectrometer. The measurements were carried out at 122 250 µA trap current and an electron accelerating voltage of 60 eV. The three Ne isotopes were 123 simultaneously detected on three discrete dynode multipliers operating in pulse counting mode.

124
Large 20 Ne beams (>100,000 cps) were measured on a Faraday cup. An automated liquid 125 nitrogen trap was used to keep the Ar and CO 2 backgrounds low and we corrected for isobaric 126 interferences from doubly-charged Ar and CO 2 . The 40 Ar ++ / 40 Ar + and CO 2 ++ /CO 2 + ratios were 127 0.031±0.003 and 0.0045±0.0005, respectively, and the 40 Ar ++ and CO 2 ++ corrections were all 128 below 1%. For Ar, depending on the abundance measured by the quadrupole mass spectrometer, 129 a fraction of the gas was let into the mass spectrometer. Isotopes were measured simultaneously 130 using the Faraday for 40 Ar and the axial and low mass multipliers for 38 Ar and 36 Ar, respectively.

131
Xe was measured using the three discrete dynode multipliers in a combination of multicollection 132 and peak jumping mode. Additional analytical details are described in Mukhopadhyay (2012 these two samples, the step crush data do in general fall on the hyperbolic best fit line for NLD 195 27 (Fig 4). Hence, all of the samples may have similar mantle source 40 Ar/ 36 Ar values.

196
The hyperbolic fit for NLD 27 in Ar-Xe space yields a mantle source value of 6.92 ± 3). As seen in Figure 3, for a given 4 He/ 3 He ratio, the Rochambeau samples have a higher does not explain the low 40 Ar/ 36 Ar ratios of the plume sources (Fig. 6a). Consequently, a less 249 degassed source is required to explain the lower 40 Ar/ 36 Ar ratio of the plume source.     Table 3). To investigate 340 whether inclusion of some of the less precise measurements affect the fission deconvolution, the 341 above analyses were redone using a filtered data set; only data points with 132 Xe/ 136 Xe distinct 342 from the atmospheric composition at the 2σ level and with a relative error of <1% were selected.

343
Such filtering only eliminates 4 data points and does not affect the deconvolution.

344
Following determination of the mantle source composition, the least-squares solution to  Table 3.

360
Depending on whether the initial mantle Xe is solar or chondritic, the fraction of 136 Xe 361 derived from 244 Pu fission is 0.87±0.11 or 0.85±0.14, respectively (  (Table 2). While detailed 381 modeling of the fission and radiogenic Xe isotopes is beyond the scope of this paper, we note 382 that the 129 Xe*/ 136 Xe Pu * values at Rochambeau are comparable to those from Iceland but 383 significantly lower than the MORB source (Table 1). Interpreting these values as closure ages for 384 a mantle with an initially homogenous I/Pu ratio, the higher 129 Xe*/ 136 Xe Pu * ratio in the depleted 385 MORB source would imply that the shallow upper mantle became closed to volatile loss prior to 386 the deep mantle reservoir supplying noble gases to the mantle plumes. Such a conclusion appears 387 paradoxical. Rather, a simpler explanation is that the lower 129 Xe*/ 136 Xe Pu * in the Rochambeau 388 and Iceland source reflects a lower initial I/Pu ratio for the plume source compared to the MORB 389 source. This difference would suggest that the initial phase of Earth's accretion was volatile poor 390 compared to the later stages of accretion because Pu is a refractory element while I is a volatile 391 element (e.g., Mukhopadhyay, 2012). Since this difference in I/Pu ratio is still preserved in the 392 present day Xe isotopic ratio of the mantle, we argue that the whole mantle was never 393 completely homogenized. MORBs. While originally the plume source was assumed to be the whole lower mantle, the basic 419 framework could still be viable if instead of the whole lower mantle, the plume source was much 420 smaller, such as D" (e.g., Tolstikhin et al., 2006).

473
Several studies have suggested that the primitive-looking 4 He/ 3 He ratios in OIBs are 474 signatures of depleted residues of mantle melting because U is more incompatible than He (e.g., 475 Coltice and Ricard, 1999; Parman et al., 2005). In such scenarios, separation of the MORB and 476 low 4 He/ 3 He reservoirs is not required over Earth's history. Rather, because the residues have 477 very low U/ 3 He ratios, the 4 He/ 3 He ratio of the convecting mantle gets frozen in the residues. For 478 residues generated at 2-3 Ga, the convecting mantle 4 He/ 3 He could have the same values as 479 observed in many OIBs. Our results from Iceland and Rochambeau suggest that if low 4 He/ 3 He 480 ratios in OIBs are indeed due to sampling of depleted residues of mantle melting, then the 481 129 Xe/ 130 Xe ratios require the depleted residues to be generated prior to 4.45 Ga. In other words, 482 the low 4 He/ 3 He reservoir has essentially behaved as a closed system over Earth's history. Rochambeau Xe data we can conclusively say that these features must have been produced prior 491 to 4.45 Ga (Figs. 6 and 7). Therefore, LLSVPs are long lasting structures in the deep mantle and 492 are essentially as old as the age of the Earth.

493
Our observation that the Rochambeau and Iceland plume sources have high proportions 494 of Pu-derived fission Xe as well as recycled atmospheric Xe requires that plumes sample both 495 primitive and recycled material. We note that the DICE 10 sample from Iceland has amongst the 496 most primitive 21 Ne/ 22 Ne ratio, yet ~90% of its Xe is from a recycled source (Table 1). Hence, if 497 all of the plume material is derived from LLSVPs then these features must also be composed of