Volume 20, Issue 12 p. 5939-5967
Research Article
Free Access

Hot and Heterogenous High-3He/4He Components: New Constraints From Proto-Iceland Plume Lavas From Baffin Island

Lori N. Willhite

Corresponding Author

Lori N. Willhite

Department of Earth Science, University of California Santa Barbara, Santa Barbara, CA, USA

Correspondence to: L. N. Willhite,

[email protected]

Search for more papers by this author
Matthew G. Jackson

Matthew G. Jackson

Department of Earth Science, University of California Santa Barbara, Santa Barbara, CA, USA

Search for more papers by this author
Janne Blichert-Toft

Janne Blichert-Toft

Laboratoire de Géologie de Lyon, Ecole Normale Supérieure de Lyon, CNRS UMR 5276, Université de Lyon, Lyon, France

Search for more papers by this author
Ilya Bindeman

Ilya Bindeman

Department of Earth Sciences, University of Oregon, Eugene, OR, USA

Search for more papers by this author
Mark D. Kurz

Mark D. Kurz

Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, MA, USA

Search for more papers by this author
Sæmundur A. Halldórsson

Sæmundur A. Halldórsson

NordVulk, Institute of Earth Sciences, University of Iceland, Reykjavík, Iceland

Search for more papers by this author
Sunna Harðardóttir

Sunna Harðardóttir

NordVulk, Institute of Earth Sciences, University of Iceland, Reykjavík, Iceland

Search for more papers by this author
Esteban Gazel

Esteban Gazel

Department of Earth and Atmospheric Sciences, Cornell University, Ithaca, NY, USA

Search for more papers by this author
Allison A. Price

Allison A. Price

Department of Earth Science, University of California Santa Barbara, Santa Barbara, CA, USA

Search for more papers by this author
Benjamin L. Byerly

Benjamin L. Byerly

Department of Earth Science, University of California Santa Barbara, Santa Barbara, CA, USA

Search for more papers by this author
First published: 07 November 2019
Citations: 15

Abstract

The Icelandic hotspot has erupted basaltic magma with the highest mantle-derived 3He/4He over a period spanning much of the Cenozoic, from the early-Cenozoic Baffin Island-West Greenland flood basalt province (49.8 RA), to mid-Miocene lavas in northwest Iceland (40.2 to 47.5 RA), to Pleistocene lavas in Iceland's neovolcanic zone (34.3 RA). The Baffin Island lavas transited through and potentially assimilated variable amounts of Precambrian continental basement. We use geochemical indicators sensitive to continental crust assimilation (Nb/Th, Ce/Pb, MgO) to identify the least crustally contaminated lavas. Four lavas, identified as “least crustally contaminated,” have high MgO (>15 wt.%), and Nb/Th and Ce/Pb that fall within the mantle range (Nb/Th = 15.6 ± 2.6, Ce/Pb = 24.3 ± 4.3). These lavas have 87Sr/86Sr = 0.703008–0.703021, 143Nd/144Nd = 0.513094–0.513128, 176Hf/177Hf = 0.283265–0.283284, 206Pb/204Pb = 17.7560–17.9375, 3He/4He up to 39.9 RA, and mantle-like δ18O of 5.03–5.21‰. The radiogenic isotopic compositions of the least crustally contaminated lavas are more geochemically depleted than Iceland high-3He/4He lavas, a shift that cannot be explained by continental crust assimilation in the Baffin suite. Thus, we argue for the presence of two geochemically distinct high-3He/4He components within the Iceland plume. Additionally, the least crustally contaminated primary melts from Baffin Island-West Greenland have higher mantle potential temperatures (1510 to 1630 °C) than Siqueiros mid-ocean ridge basalts (1300 to 1410 °C), which attests to a hot, buoyant plume origin for early Iceland plume lavas. These observations support the contention that the geochemically heterogeneous high-3He/4He domain is dense, located in the deep mantle, and sampled by only the hottest plumes.

Key Points

  • Baffin Island-West Greenland high-3He/4He lavas are more geochemically depleted than any other high-3He/4He lavas globally
  • The isotopic composition of the high-3He/4He mantle source in the Iceland plume has evolved through time
  • Baffin Island and West Greenland primary melts record hotter temperatures than high-MgO MORB, consistent with a deep, dense plume source

1 Introduction

Helium isotopes provide an important tracer of ancient domains that have survived inside the Earth since its accretion. Helium isotopic ratios (normalized to Earth's atmosphere, 3He/4He = 1.384 × 10-6) are relatively constant in mid-ocean ridge basalts, or MORB, (8.8 ± 2.1 RA, or ratio to atmosphere; Graham, 2002), which passively sample the upper mantle. However, plume-fed hotspots—such as Iceland, Hawaii, Samoa, and Galápagos—sample mantle domains with much higher 3He/4He ratios, thought to be located in the deep mantle (>30 RA; e.g., Ellam & Stuart, 2004; Farley et al., 1992; Hilton et al., 1999; Jackson, Hart, et al., 2007; Kurz et al., 1982; Macpherson et al., 2005; Saal et al., 2007; Starkey et al., 2009). The highest observed mantle-derived 3He/4He (up to 49.8 ± 0.7 RA) was found in the continental flood basalts associated with the Iceland plume at Baffin Island and West Greenland, erupted at ~60 Ma (Storey et al., 1998; Rizo et al., 2016; Starkey et al., 2009; Stuart et al., 2003; Graham et al., 1998). Elevated 3He/4He ratios were also identified in lavas related to the Iceland plume in east Greenland (Marty et al., 1998). Mid-Miocene lavas in northwest Iceland host the highest observed mantle-derived 3He/4He of any ocean island basalt (OIB) location (47.5 RA, Harðardóttir et al., 2018; 40.2 RA, Mundl et al., 2017; 37.7 RA, Hilton et al., 1999). Modern Iceland lavas in the neovolcanic zone also have high 3He/4He (up to 34.3 RA; Macpherson et al., 2005). Therefore, the Iceland plume has hosted elevated 3He/4He over much of its history and, hence, is an ideal natural laboratory for studying the high-3He/4He mantle domain.

The high-3He/4He mantle domain is ancient, requiring preservation in a region of the mantle that is relatively undegassed despite billions of years of mantle convective mixing, melting, and recycling (e.g., Class & Goldstein, 2005; Tackley, 2000; White, 2015; Zindler & Hart, 1986). Therefore, constraining the composition of the highest 3He/4He mantle reservoir observed in the rock record can provide important new insights into the accretionary history and early evolution of Earth's major chemical reservoirs. This study examines the geochemistry of flood basalts from Baffin Island, Canada (Figure 1), and West Greenland, providing new data—He-O-Sr-Nd-Hf-Pb isotopic compositions, as well as whole rock major and trace element concentrations—for 18 lavas from Baffin Island, in order to constrain the composition of the mantle domain with the highest observed 3He/4He.

Details are in the caption following the image
Map of Baffin Island, Greenland, and Iceland. General locations of Iceland plume-derived lavas are shaded in dark gray. The inset shows a simplified geologic map after Wheeler et al. (1996), including the locations of the lavas collected in this study: Padloping Island, Durban Island, and Akpat Point. The hotspot track is a synthetic track with the North American plate fixed through time (modified after Lawver & Müller, 1994) and is shown as the gray path. Also shown is the location of the ~60 Ma West Greenland succession samples compiled in Larsen and Pedersen (2009).

The Baffin Island and West Greenland lavas constitute a flood basalt province associated with the proto-Icelandic plume that erupted through Archean and Proterozoic continental crust, the assimilation of which could have overprinted their primary mantle signature (e.g., Day, 2016; Saunders et al., 1997). Therefore, we identify signatures of crustal assimilation in Baffin Island and West Greenland lavas using a suite of major and trace element filters—whole rock MgO, Ce/Pb, and Nb/Th—sensitive to continental crust assimilation, in order to isolate geochemical signatures of their mantle source. We show that, among high-3He/4He lavas globally, the least contaminated lavas from Baffin Island have the most geochemically depleted 87Sr/86Sr, 143Nd/144Nd, and 176Hf/177Hf, and the least radiogenic Pb isotopic compositions. Baffin Island-West Greenland lavas exhibit more geochemically depleted isotopic fingerprints than the high-3He/4He lavas erupted in mainland Iceland, demonstrating temporal evolution of the high-3He/4He component in the Iceland hotspot. The observation of two geochemically distinct, high-3He/4He components in a single hotspot provides new constraints on the origin and evolution of mantle domains hosting high 3He/4He.

2 Methods

2.1 Rock Collection, Preparation, and Analytical Methods

The 18 basalts examined in this study were collected at three locations on Baffin Island by Don Francis during the 2004 field season—Padloping Island, Akpat Point, and Durban Island (Figure 1 and Table S1 in the supporting information). Eleven of the 18 samples in this study have fresh volcanic glass on the margins of the basaltic pillows, a feature that has been identified previously in Baffin Island flood basalt lavas (e.g., Kent et al., 2004). The supporting information methods describe sample preparation and analytical techniques for whole rock major and trace element analyses (Table 1), whole rock isotopic analyses (He, Sr, Nd, Hf, Pb; Table 2), oxygen isotopic analyses of olivines (Table 3), and olivine compositions (supporting information Table S2). Thin section images of all 18 samples are provided in the supporting information Figure S1.

Table 1. Major and Trace Elements of Baffin Island Lavas and Reference Materialsa
AK-1 AK-6 AK-8b AK-9 AK-12 AK-13 AK-14 AK-18A DB-9 DB-13 DB-14 DB-17 DB-19 PI-10 PI-15 PI-17 PI-18 PI-20 BCR-2 BCR-2 BCR-2 publ BHVO-2 BHVO-2 BHVO-2 publ
XRF
SiO2 (wt.%) 45.05 43.80 45.83 45.40 45.30 44.88 44.87 44.81 44.16 44.69 45.10 46.52 46.11 46.26 44.59 46.28 45.73 45.59 53.96 54.46 54.93 50.04 50.23
TiO2 0.69 0.54 0.96 0.52 0.69 0.69 0.68 0.70 0.69 0.76 0.77 0.70 1.29 1.00 0.75 0.86 0.84 0.83 2.28 2.30 2.30 2.78 2.77
Al2O3 11.54 8.63 11.04 9.83 11.51 11.61 11.47 11.76 9.86 10.95 10.92 11.60 11.38 12.85 10.60 12.18 11.83 11.61 13.56 13.62 13.71 13.68 13.61
FeOT 10.43 10.57 10.88 10.51 10.37 10.60 10.46 10.47 10.53 10.82 10.76 10.13 10.79 10.75 10.66 10.43 10.59 10.55 12.85 12.62 12.61 11.25 11.29
MnO 0.17 0.17 0.18 0.17 0.17 0.18 0.17 0.18 0.17 0.18 0.18 0.17 0.18 0.18 0.17 0.17 0.17 0.17 0.20 0.20 0.20 0.17 0.17
MgO 19.31 24.28 19.85 23.63 19.40 19.53 19.65 19.43 23.01 20.81 20.76 18.90 17.03 15.17 21.58 17.73 18.48 18.56 3.59 3.53 3.66 7.30 7.35
CaO 10.25 7.27 9.41 8.21 10.51 10.07 10.28 10.04 8.26 9.09 9.06 9.76 9.23 10.86 8.94 10.08 9.77 9.62 7.15 7.16 7.24 11.51 11.54
Na2O 1.22 0.88 1.30 0.98 1.26 1.27 1.15 1.26 0.88 1.09 1.04 1.23 1.39 1.35 1.08 1.31 1.26 1.21 3.17 3.10 3.17 2.24 2.25
K2O 0.02 0.06 0.04 0.01 0.02 0.02 0.01 0.04 0.05 0.01 0.02 0.01 0.17 0.02 0.01 0.04 0.04 0.05 1.79 1.79 1.80 0.52 0.52
P2O5 0.05 0.04 0.07 0.05 0.05 0.05 0.04 0.04 0.06 0.05 0.05 0.05 0.12 0.07 0.06 0.07 0.07 0.07 0.35 0.35 0.37 0.26 0.27
Total (majors only)b 98.74 96.24 99.55 99.32 99.29 98.88 98.79 98.71 97.67 98.46 98.64 99.08 97.71 98.50 98.44 99.16 98.79 98.26 98.92 99.13 100.00 99.74 100.00
LOI 0.22 3.27 0.00 0.13 0.32 0.00 0.55 0.59 1.37 0.47 0.65 0.05 1.18 0.49 0.43 0.56 0.45 0.55 0.10 0.00 0.00
Total (major, trace oxides, LOI)b 99.38 100.05 99.96 99.96 100.03 99.30 99.76 99.72 99.52 99.37 99.73 99.54 99.28 99.30 99.32 100.10 99.64 99.21 99.27 99.39 99.99
Olivine Fo# c 87.0 83.7 89.6 89.0 87.2 87.1 87.6 87.3 88.0 87.5 87.0 88.8 86.4 87.1 88.7 90.4 89.9 90.4
XRF
Rb (ppm) 0.97 1.86 1.18 1.96 1.18 1.26 1.36 0.59 2.35 1.07 1.17 1.10 3.38 1.97 1.07 0.97 1.56 1.84 47 47 46 11 9
Sr 63.8 44.7 113.3 47.4 61.5 63.6 60.9 65.4 70.7 77.6 76.9 75.1 170.9 109.0 74.3 99.8 94.8 92.7 340 343 337 397 394
Zn 82.0 70.2 75.8 74.0 76.8 76.3 71.8 74.3 82.3 79.4 78.6 73.1 81.3 77.9 85.9 74.1 77.9 77.1 133 130 130 105 104
Ni 779 1094 849 1100 789 790 790 778 1038 878 871 763 638 486 943 684 734 738 14 13 13 121 120
Cr 1605 2274 1398 2054 1630 1652 1648 1616 1805 1697 1711 1591 1304 1016 1720 1415 1486 1470 9 12 16 283 287
V 241 197 248 199 240 239 236 243 212 231 232 235 288 272 228 245 241 235 405 408 418 321 318
Cu 108.0 77.7 96.4 68.7 113.7 110.8 110.8 113.3 97.5 97.1 87.5 114.2 79.7 119.1 86.2 93.0 83.9 87.6 20 19 20 130 129
Ga 11.4 8.3 11.5 9.7 11.5 11.9 12.5 11.4 10.5 12.0 10.3 12.9 12.3 14.0 11.3 12.5 10.8 11.2 20 22 22 21 21
Ba 9.2 8.2 13.7 9.8 9.6 7.5 8.2 5.5 10.9 4.9 9.5 12.0 74.8 23.0 6.4 18.2 22.6 22.1 676 678 684 139 131
Y 16.8 11.9 15.9 12.3 16.2 15.7 15.4 16.7 13.4 16.5 15.2 15.6 24.7 18.6 14.8 17.5 15.8 16.7 36 36 36 26 26
Nb 0.50 0.10 1.00 0.00 0.60 0.20 0.40 1.19 0.60 0.89 0.10 0.96 7.53 3.15 0.00 4.55 2.59 3.76 12 13 12 16 18
Zr 36.4 28.9 53.1 26.0 36.8 36.9 35.4 36.3 39.2 42.4 41.1 36.6 80.8 54.2 39.8 50.0 49.1 49.4 182 181 187 168 171
ICP-MS
Cs (ppm) 0.0027 0.014 0.0020 0.0004 0.0058 0.0011 0.0044 0.0043 0.0081 0.0029 0.0034 0.0055 0.0070 0.0041 0.0032 0.0060 0.0055 0.0093 1.13 1.16 0.098 0.100
Rb 0.26 1.08 0.28 0.15 0.30 0.20 0.23 0.42 0.95 0.12 0.17 0.26 2.95 0.24 0.18 0.31 0.41 0.61 46.3 46.0 8.95 9.26
Ba 6.04 6.21 11.18 4.77 5.76 4.87 5.84 7.36 7.32 6.55 7.89 9.60 71.37 15.20 5.97 21.63 22.76 25.91 679 684 130 131
Th 0.194 0.154 0.128 0.040 0.193 0.183 0.188 0.183 0.111 0.088 0.086 0.112 0.713 0.294 0.098 0.416 0.410 0.392 6.18 5.83 1.26 1.22
U 0.044 0.037 0.035 0.012 0.044 0.040 0.044 0.049 0.029 0.014 0.016 0.028 0.146 0.043 0.018 0.054 0.049 0.063 1.54 1.68 0.41 0.41
Nb 1.12 0.95 1.73 0.51 1.07 1.08 1.08 1.08 1.39 1.17 1.17 1.15 7.85 3.62 1.36 4.26 4.16 4.12 12.6 12.4 19.4 18.1
Ta 0.069 0.061 0.112 0.034 0.067 0.066 0.070 0.068 0.089 0.080 0.076 0.069 0.460 0.211 0.087 0.250 0.242 0.239 0.81 0.79 1.27 1.15
La 1.59 1.29 2.32 0.70 1.54 1.52 1.51 1.56 1.68 1.47 1.32 1.58 6.56 3.51 1.63 3.67 3.18 3.25 26.1 25.1 15.7 15.2
Ce 4.23 3.35 6.43 2.18 4.07 4.10 4.04 4.15 4.63 4.37 4.30 4.02 14.12 8.48 4.69 8.35 8.01 7.93 51.1 53.1 36.3 37.5
Pb 0.35 0.22 0.28 0.15 0.33 0.34 0.34 0.37 0.19 0.20 0.20 0.21 0.80 0.38 0.19 0.36 0.34 0.34 10.0 10.6 1.58 1.65
Pr 0.69 0.54 1.09 0.42 0.67 0.68 0.67 0.69 0.78 0.77 0.73 0.70 1.99 1.31 0.79 1.22 1.10 1.13 6.57 6.83 5.17 5.34
Nd 3.72 2.93 5.71 2.42 3.68 3.65 3.64 3.69 4.03 4.19 3.99 3.79 9.18 6.35 4.32 5.92 5.36 5.51 27.0 28.3 23.3 24.3
Sr 64.4 44.4 110.2 48.9 60.2 62.3 60.4 65.7 70.7 77.1 75.7 76.7 170.7 110.1 74.2 100.7 92.3 91.3 347 337 396 394
Zr 33.9 25.8 50.0 23.8 32.6 33.0 32.8 33.7 36.1 38.5 38.3 34.1 80.1 53.7 38.2 48.9 46.8 46.9 188 187 173 171
Hf 0.99 0.76 1.38 0.68 0.96 0.97 0.97 0.98 0.98 1.11 1.08 0.94 2.06 1.47 1.05 1.31 1.28 1.29 4.81 4.97 4.40 4.47
Sm 1.54 1.18 2.10 1.11 1.52 1.51 1.52 1.48 1.54 1.69 1.61 1.48 3.04 2.26 1.64 2.05 1.87 1.89 7.01 6.55 6.45 6.02
Eu 0.65 0.48 0.83 0.46 0.62 0.63 0.63 0.64 0.60 0.66 0.65 0.62 1.08 0.89 0.64 0.78 0.73 0.74 2.15 1.99 2.24 2.04
Gd 2.33 1.72 2.69 1.71 2.23 2.27 2.22 2.28 2.09 2.31 2.27 2.22 3.84 2.96 2.25 2.69 2.54 2.52 7.11 6.81 6.64 6.21
Tb 0.46 0.33 0.49 0.34 0.43 0.44 0.43 0.44 0.39 0.44 0.42 0.41 0.69 0.53 0.42 0.50 0.47 0.48 1.17 1.08 1.04 0.94
Dy 3.02 2.23 3.15 2.30 2.93 2.97 2.92 3.00 2.53 2.87 2.81 2.78 4.45 3.49 2.74 3.27 3.07 3.10 7.09 6.42 5.87 5.28
Ho 0.65 0.49 0.64 0.50 0.63 0.63 0.63 0.65 0.54 0.61 0.60 0.60 0.95 0.74 0.58 0.69 0.65 0.66 1.42 1.31 1.08 0.99
Y 16.2 12.0 16.0 12.8 15.7 15.8 15.6 16.2 13.5 15.1 14.6 15.0 23.5 18.4 14.6 17.0 16.0 16.2 35.7 36.1 25.9 25.9
Er 1.82 1.39 1.73 1.43 1.77 1.79 1.80 1.80 1.50 1.71 1.66 1.69 2.62 2.04 1.64 1.93 1.81 1.84 3.87 3.67 2.67 2.51
Tm 0.26 0.20 0.24 0.20 0.25 0.26 0.25 0.26 0.22 0.24 0.24 0.25 0.37 0.30 0.23 0.27 0.26 0.26 0.55 0.53 0.34 0.33
Yb 1.66 1.23 1.49 1.33 1.61 1.62 1.60 1.65 1.34 1.53 1.50 1.52 2.35 1.80 1.45 1.71 1.64 1.65 3.38 3.39 2.01 1.99
Lu 0.26 0.19 0.23 0.21 0.25 0.25 0.24 0.25 0.21 0.24 0.24 0.23 0.36 0.28 0.24 0.27 0.25 0.26 0.52 0.505 0.29 0.28
Sc 36.1 28.9 33.0 31.3 34.4 35.4 34.4 36.1 31.2 34.0 34.0 34.5 34.3 38.9 33.4 37.0 35.5 35.7 34.3 33.5 31.7 31.8
Ba/Th 31.2 40.4 87.1 119.8 29.8 26.5 31.0 40.1 65.8 74.5 91.4 85.9 100.1 51.6 60.8 52.0 55.5 66.2 110 117 103 107
Ce/Pb 12.1 15.5 22.9 14.7 12.3 12.2 11.8 11.2 24.0 22.2 21.7 19.4 17.6 22.4 24.1 22.9 23.8 23.3 5.10 5.02 22.99 22.70
Nb/U 25.6 25.8 50.0 42.1 24.3 26.8 24.7 21.8 47.2 86.4 74.4 40.6 53.7 84.0 77.8 78.3 85.3 64.9 8.14 7.39 47.05 43.93
Nb/Th 5.8 6.2 13.5 12.8 5.5 5.9 5.7 5.9 12.5 13.3 13.5 10.3 11.0 12.3 13.9 10.3 10.2 10.5 2.04 2.13 15.42 14.79
[La/Sm]N 0.65 0.68 0.69 0.39 0.63 0.63 0.62 0.66 0.68 0.55 0.52 0.67 1.35 0.98 0.62 1.13 1.06 1.08 2.33 2.40 1.53 1.58
Rb/Cs 95.8 78.1 134.9 359.9 52.1 188.6 52.3 97.6 116.6 42.1 48.7 46.9 424.3 58.1 54.8 52.1 74.1 65.7 41.0 39.7 91.5 93.0
Ba/Rb 23.6 5.8 40.7 31.8 19.0 24.5 25.2 17.7 7.7 53.2 47.6 37.0 24.2 63.6 33.8 69.0 56.2 42.6 14.7 14.9 14.5 14.1
Th/U 4.4 4.2 3.7 3.3 4.4 4.5 4.3 3.7 3.8 6.5 5.5 3.9 4.9 6.8 5.6 7.6 8.4 6.2 4.0 3.5 3.1 3.0
  • a Majors and some traces were analyzed by XRF at WSU. The other traces were analyzed by ICP-MS at WSU. Two USGS reference materials, BCR-2 and BHVO-2, were run together with the Baffin lavas as unknowns. These data are provided with preferred values from Jochum et al. (2016) (data are expressed with all Fe as FeO to facilitate comparison with new BCR-2 and BHVO-2 provided here).
  • b Two different totals are incluced for major element analyses. The first total includes major element analyses only. The second total includes major element analyses, LOI (loss on ignition), and the trace element totals expressed as oxides (and includes the following trace elements: Ni, Cr, Sc, V, Ba, Rb, Sr, Zr, Y, Nb, Ga, Cu, Zn, Pb, La, Ce, Th, Nd and U).
  • c Olivine forserite compositions are average values of multiple analyses of different olivine grains provided in Supporting Information Table S2.
Table 2. New Sr, Nd, Hf, Pb, He, and O Isotopic Compositions of Baffin Island Lavas
Sample name Location Sample typea 87Sr/86Sr 2 σ 143Nd/144Nd 2 σ ɛ143Ndb 176Hf/177Hf 2 σ 206Pb/204Pb 2 σ 207Pb/204Pb 2 σ 208Pb/204Pb 2 σ 207Pb/206Pb 2 σ 208Pb/206Pb 2 σ δ18O oliv 3He/4Hec 1 σ 4He cc STP/g Olivine mass (g) Fraction He blank
AK-1 Akpat Pt. Glass 0.703559 0.000006 0.512963 0.000006 6.5 0.283231 0.000004 17.6822 0.0013 15.2945 0.0014 37.751 0.003 0.86496 0.00002 2.13501 0.00006 5.21 1.50 0.08 2.76E-09 0.18390 0.06
AK-6 Akpat Pt. Rock chips 0.703501 0.000006 0.512997 0.000003 7.2 0.283222 0.000005 17.6249 0.0009 15.2887 0.0010 37.664 0.003 0.86747 0.00002 2.13704 0.00008 5.33 2.9 0.8 6.15E-11 0.20507 0.70
AK-8b Akpat Pt. Rock chips 0.703009 0.000006 0.513128 0.000003 9.7 0.283266 0.000003 17.7560 0.0010 15.3932 0.0009 37.532 0.002 0.86694 0.00001 2.11373 0.00004 5.03 39.9 0.5 2.30E-08 0.18252 0.01
AK-8b fusion 20.8 0.5 1.84E-08 0.16757 0.01
AK-9 Akpat Pt. Rock chips 0.702995 0.000006 0.513174 0.000003 10.6 0.283287 0.000005 17.7715 0.0050 15.3812 0.0045 37.500 0.012 0.86557 0.00003 2.11018 0.00006 56.6 1.1 1.42E-10 0.27417 0.43
AK-9 fusion 36.3 0.9 5.23E-09 0.25194 0.01
AK-12 Akpat Pt. Glass 0.703579 0.000007 0.512954 0.000006 6.3 0.283234 0.000004 17.6890 0.0015 15.2932 0.0015 37.738 0.004 0.86458 0.00003 2.13340 0.00006 30.1 0.7 2.77E-10 0.18074 0.37
AK-13d Akpat Pt. Glass 0.703579 0.000008 0.512957 0.000005 6.4 0.283212 0.000005 17.6601 0.0018 15.2930 0.0018 37.700 0.005 0.86600 0.00002 2.13480 0.00007 5.32 28.8 0.4 7.83E-09 0.07340 0.05
AK-13 crush replicated 21.5 0.5 7.07E-10 0.13692 0.24
AK-14 Akpat Pt. Glass 0.703618 0.000021 0.512956 0.000006 6.4 0.283218 0.000005 17.6951 0.0028 15.3013 0.0025 37.762 0.006 0.86470 0.00003 2.13396 0.00008 5.12 17.9 0.4 5.65E-10 0.19229 0.22
AK-18a Akpat Pt. Glass 0.703635 0.000006 0.512952 0.000006 6.3 0.283229 0.000004 17.7029 0.0045 15.3128 0.0040 37.761 0.010 0.86499 0.00003 2.13315 0.00006 15.4 0.4 3.65E-10 0.20209 0.29
DB-9 Durban Is. Rock chips 0.702997 0.000009 0.513135 0.000003 9.8 0.283272 0.000004 17.9507 0.0031 15.4168 0.0030 37.717 0.008 0.85886 0.00004 2.10123 0.00010 5.10 13.3 0.7 9.27E-11 0.22594 0.59
DB-13 Durban Is. Glass 0.703021 0.000005 0.513102 0.000003 9.2 0.283265 0.000004 17.9317 0.0025 15.4291 0.0021 37.732 0.006 0.86044 0.00002 2.10434 0.00015 5.03 12.0 0.7 6.75E-11 0.19448 0.70
DB-14 Durban Is. Glass 0.703021 0.000005 0.513097 0.000003 9.1 0.283284 0.000003 17.9297 0.0025 15.4279 0.0030 37.735 0.007 0.86047 0.00004 2.10453 0.00008 5.09 10.1 0.5 6.56E-11 0.21668 0.68
DB-17 Durban Is. Rock chips 0.703228 0.000006 0.513104 0.000003 9.2 0.283230 0.000006 17.5114 0.0042 15.2942 0.0037 37.455 0.009 0.87341 0.00002 2.13882 0.00007 5.17 31.2 0.9 7.53E-11 0.20028 0.66
DB-17 fusion 6.4 0.4 4.68E-09 0.18434 0.01
DB-19 Durban Is. Rock chips 0.703946 0.000005 0.512937 0.000003 6.0 0.283144 0.000004 18.0095 0.0012 15.3929 0.0009 37.971 0.003 0.85472 0.00002 2.10846 0.00005 5.08 0.2 0.4 1.14E-10 0.19551 0.50
PI-10 Padloping Is. Rock chips 0.703401 0.000005 0.513028 0.000003 7.8 0.283222 0.000004 17.9607 0.0032 15.4001 0.0035 37.920 0.011 0.85735 0.00003 2.11085 0.00004
PI-15 Padloping Is. Glass 0.703008 0.000005 0.513094 0.000003 9.0 0.283279 0.000003 17.9375 0.0029 15.4223 0.0025 37.723 0.008 0.85980 0.00002 2.10296 0.00006 5.21 21.8 0.6 1.93E-10 0.20635 0.42
PI-17 Padloping Is. Glass 0.703845 0.000006 0.512926 0.000003 5.8 0.283169 0.000004 17.7551 0.0013 15.3663 0.0012 37.662 0.002 0.86545 0.00001 2.12113 0.00015 36.9 0.5 2.42E-09 0.23348 0.05
PI-18 Padloping Is. Glass 0.703848 0.000006 0.512920 0.000003 5.7 0.283169 0.000004 17.7542 0.0012 15.3680 0.0011 37.660 0.003 0.86560 0.00002 2.12116 0.00004 5.18 36.4 0.6 1.23E-09 0.19692 0.11
PI-20 Padloping Is. Glass 0.703846 0.000006 0.512923 0.000003 5.7 0.283182 0.000003 17.7540 0.0015 15.3642 0.0015 37.659 0.004 0.86540 0.00002 2.12107 0.00007 5.07 31.1 0.7 7.60E-10 0.23062 0.15
BCR-2e 0.512624 0.000003
BCR-2e 0.512621 0.000004
BCR-2e 0.512627 0.000004
BCR-2e, f 0.705000 0.000005 0.512621 0.000003
AGV-2f 0.703966 0.000006
AGV-2f 0.703966 0.000005
AGV-2f 0.703972 0.000006
BCR-2g 0.282886 0.000003 18.7558 0.0009 15.6251 0.0008 38.7414 0.0022
AGV-2g 0.282988 0.000006 18.8660 0.0009 15.6174 0.0009 38.5299 0.0032
  • a For the 11 samples with available pillow glass, heavy radiogenic isotopies were measured on glass; for the remaining samples they were measured on rock chips.
  • b ɛ143Nd is calculated assuming a chondritic value from Bouvier et al. (2008).
  • c With three exceptions, all of the helium isotopic analyses were made following crushing olivines in vacuo. Using the crushed powders remaining for the 3He/4He crush analyses, three fusion experiments were conducted here, on samples DB-17, AK-8b, and AK-9.
  • d For AK-13, the 28.8 RA olivine crush values is for a single megacryst (73 mg), and the 21.5 RA replicate value is from 137 mg of smaller olivine crystals.
  • e An aliquot of the USGS reference material BCR-2 was run with each analytical session for Nd isotopic analyses.
  • f Aliquots of at least one of two USGS reference materials, AGV-2 and BCR-2, were run with each analytical session for Sr isotopic analyses.
  • g Hf and Pb isotopic compositions of the USGS reference materials BCR-2 and AGV-2 shown here were reported in Price et al. (2016) and were run in separate analytical sessions from the samples in this study, but in the same laboratory (Lyon) following exactly the same procedures (described in Price et al., 2016).
Table 3. Oxygen Isotopic Compositions of Baffin Island Olivines
Sample ID Mass (mg) δ18O μmol O2 μmols O2/mg
AK-1 1.48 5.24 0.07 19.11 12.54
AK-1-replicate 1.43 5.19 0.07 18.79 12.76
sample average 5.21
AK-6 1.05 5.38 0.07 14.24 13.56
AK-6-replicate 1.20 5.27 0.07 15.38 12.82
AK-6-rep2 0.70 5.35 0.07 9.1 13.00
sample average 5.33
AK-8b 1.36 5.08 0.07 18.86 13.87
AK-8b-replicate 1.67 4.98 0.07 23.18 13.88
sample average 5.03
AK-13 1.29 5.24 0.07 17.28 13.40
AK-13-replicate 0.86 5.40 0.10 11.1 12.98
sample average 5.32
AK-14 0.90 5.12 0.07 11.27 12.16
DB-9 1.32 5.06 0.07 17.45 13.22
DB-9-replicate 1.18 5.14 0.07 15.12 12.81
sample average 5.10
DB-13 1.40 5.03 0.07 18.36 12.73
DB-14 1.22 5.10 0.07 16.31 13.37
DB-14-replicate 1.25 5.09 0.07 16.77 13.42
sample average 5.09
DB-17 1.33 5.16 0.07 16.03 12.05
DB-17-replicate 1.21 5.18 0.07 16.78 13.87
sample average 5.17
DB-19 1.19 5.02 0.07 15.77 13.25
DB-19-replicate 1.23 5.13 0.07 16.29 13.24
sample average 5.08
PI-15 1.32 5.12 0.07 18.09 13.70
PI-15-replicate 1.27 5.31 0.07 16.72 13.17
sample average 5.21
PI-18 1.35 5.17 0.07 18.55 13.74
PI-18-replicate 1.25 5.19 0.07 17.36 13.89
sample average 5.18
PI-20 1.47 4.89 0.07 20.12 13.29
PI-20-replicate 1.64 5.17 0.07 22.23 13.16
PI-20-replicate 1.29 5.15 0.07 17.9 13.88
sample average 5.07

Additionally, in supporting information methods section S2.1, a set of criteria are established for filtering Baffin Island and West Greenland lavas that have experienced crustal assimilation. In short, lavas are considered to have experienced crustal assimilation if they have MgO < 10 wt.% and/or Ce/Pb and/or Nb/Th ratios below values representative of the mantle (i.e., Ce/Pb < 20 and Nb/Th < 13).

3 Data and Results

3.1 Major Element and Primary Melt Compositions

Major element concentrations for the Baffin Island lavas in this study are shown in Figures 2 and 3 with previously analyzed Baffin Island and West Greenland flood basalt lavas. The lavas are visually fresh and have LOI (loss on ignition) < 1.4 wt.%, with the exception of one lava with LOI = 3.3 wt.% (Table 2). The new suite of lavas reported here are tholeiitic picrites with MgO contents ranging from 15–25 wt.% (Figure 2) (Le Bas et al., 1986; Francis, 1985). To illustrate differences in major element compositions between MORB and the Baffin Island-West Greenland flood basalts, we calculated primary melt compositions of the least contaminated and least evolved (only MgO > 10 wt.% are considered) Baffin Island and West Greenland lavas, as well as high-MgO (>10 wt.%) Siqueiros MORB. Relatively few (N = 9) Baffin Island-West Greenland lavas remain after filtering for continental crust assimilation using the criteria established in section S2.1 in the supporting information. Primary melts are calculated for Siquieros, Baffin Island, and West Greenland lavas using the PRIMELT3 software assuming an Fe2O3/TiO2 ratio of 0.5 (Herzberg & Asimow, 2015). Relative to MORB, the least contaminated Baffin Island-West Greenland primary melts have higher FeO, but lower Na2O and Al2O3, and generally lower CaO and SiO2. TiO2 is not different between the two groups (Figure 3). The Baffin Island-West Greenland primary melts are highly magnesian, with calculated primary melt MgO ranging from 19–24 wt.%, which exceeds the range of calculated MgO (11–17 wt.%) in the MORB primary melts.

Details are in the caption following the image
Major element compositions of Baffin Island (BI) lavas from this study (red squares) and other sources (red diamonds; Francis, 1985; Jackson et al., 2010, Starkey et al., 2009, Stuart et al., 2003, Yaxley et al., 2004). Also shown are West Greenland (WG) lavas (also red diamonds) compiled in Larsen and Pedersen (2009). Baffin Island and West Greenland lavas are not filtered for crustal assimilation; however, only lavas with MgO > 10 wt.% are shown. High-MgO (> 10 wt.%) Siqueiros MORB samples compiled in Hays (2004) and Perfit et al. (1996) are shown for comparison. Major element compositions for Baffin Island and West Greenland differ systematically from MORB such that, for example, the flood basalt lavas have higher FeO at a given MgO. These important petrologic differences are interpreted to be the result of deeper melting and higher melt fraction in the hotter plume setting.
Details are in the caption following the image
Histogram of calculated primary melt compositions for high-MgO (>10 wt.%) Siqueiros MORB (blue) from the Siqueiros transform fault from Perfit et al. (1996) and Hays (2004) and primary melt compositions for Baffin Island (BI) and West Greenland (WG) lavas (red). The data in the histograms are consistent with hotter, deeper melting at BI-WG compared to MORB. BI-WG samples have been filtered for crustal assimilation so that all lavas plotted have Nb/Th > 13, Ce/Pb > 20, and MgO > 10 wt.%; high-Ba/Th (>100) samples from West Greenland are not considered as they are deemed modified by mantle metasomatism. Primary melts are calculated using PRIMELT3 from Herzberg and Asimow (2015) using Fe2O3/TiO2 = 0.5 and accumulated fractional melting. Mantle potential temperatures calculated with PRIMELT3 software are also shown. The number of samples for BI-WG (N = 9) is greater here than in the isotope plots because isotopes are not required for petrologic analyses.

3.2 Olivine Major and Minor Element Compositions

Olivine forsterite content from the Baffin Island lavas examined here range from forsterite 79.3 to 92.8 for individual spot analyses of at least 10 grains from each of 18 different rock samples (Figure 4 and Table S2). High forsterite olivines in Baffin Island lavas were previously reported by Francis (1985), Yaxley et al. (2004), and Starkey et al. (2012), who found forsterite compositions up to 93.2, 92.9, and 93.0, respectively. Olivines in this suite of Baffin Island lavas have higher CaO for a given forsterite than olivines found in global mantle xenoliths (Hervig et al., 1986) and mantle xenoliths from Ubekendt Ejland, West Greenland, which sample the mantle beneath the Baffin Island-West Greenland flood basalts (Bernstein et al., 2006). The CaO content of olivine reflects equilibration temperature and pressure conditions (Köhler & Brey, 1990); the high olivine CaO for a given forsterite is typical of high-temperature, low-pressure magmatic olivine and suggests that the olivines in the Baffin Island lavas are likely to be magmatic in origin (e.g., Jackson & Shirey, 2011).

Details are in the caption following the image
Olivine CaO composition compared to forsterite content of olivines in all 18 samples examined in this study. Color coding reflects maximum forsterite content: Red reflects samples with highest maximum forsterite. The CaO at a given forsterite value is distinctly higher in Baffin Island lava olivines compared to olivines found in global mantle xenoliths (from Hervig et al., 1986) and local mantle xenoliths from Ubekendt Ejland, West Greenland (Bernstein et al., 2006). Higher CaO in the picrite olivines demonstrates that these olivines were not mechanically entrained from the lithospheric mantle during magma ascent.

3.3 Trace Element Compositions

Primitive mantle-normalized (McDonough & Sun, 1995) trace element patterns, or spidergrams, are shown in Figure 5 for the Baffin Island lavas. While one sample (DB-19) has a slightly enriched rare earth element (REE) pattern ([La/Sm]N = 1.35, where N denotes normalization to primitive mantle), four lavas (PI-10, PI-17, PI-18, and PI-20) have relatively flat light REE patterns ([La/Sm]N = 0.98–1.13) and the remaining lavas have depleted light REE patterns ([La/Sm]N = 0.39–0.70). Some of the relatively fluid mobile incompatible trace elements exhibit depletions in the lavas relative to elements of similar incompatibility during mantle melting, including Cs, Rb, K, and Pb, and in some cases U. Depletions in Pb are common in mantle-derived lavas (Hart & Gaetani, 2006) and reflect either the mantle source or residual sulfide. In contrast, depletion in U and alkalis may reflect loss of these elements during subaerial weathering. For example, 13 lavas exhibit Th/U greater than the chondritic primitive mantle composition (3.876 ± 0.016; Blichert-Toft et al., 2010), with one value as high as 8.4, which likely reflects loss of U relative to immobile Th during weathering. Evidence for alkali mobility is supported by departure of Baffin Island lavas from the canonical Ba/Rb (~12; Hofmann & White, 1983) and Rb/Cs (85–95; Hofmann & White, 1983) ratios of fresh basalts. In the new suite of Baffin Island lavas, Ba/Rb and Rb/Cs vary from 5.8–69.0 and 42–424, respectively (Table 1).

Details are in the caption following the image
Primitive mantle (McDonough & Sun, 1995) normalized trace element patterns for Baffin Island lavas examined in this study plotted with an average N-MORB composition from Gale et al. (2013; using the MORB average that excludes back arc basins and lavas located <500 km from known hotspots). For elements that have both XRF and ICP-MS analyses in Table 1, ICP-MS data are plotted here.

As previously noted, large degrees of crustal assimilation are associated with low MgO, Nb/Th, and Ce/Pb in mantle-derived lavas erupted in continental settings. In Figure 6, West Greenland basement samples (compiled in Larsen & Pedersen, 2009) are shown together with Baffin Island-West Greenland lavas. At lower Nb/Th and Ce/Pb, a subset of Baffin Island-West Greenland lavas trend away from MORB-like compositions toward compositions identified in the basement.

Details are in the caption following the image
Nb/Th and Ce/Pb plotted against (top) Nb and (bottom) Ce concentrations, respectively. The least contaminated lavas from this study (N = 4) are denoted by a small black circle within the red square symbol. Continental crust rocks from West Greenland (WG; Larsen & Pedersen, 2009) have low Nb/Th and Ce/Pb. Low Nb/Th and Ce/Pb in Baffin Island and West Greenland lavas therefore are associated with higher degrees of continental crust assimilation. Baffin Island (BI) and West Greenland samples considered to be crustally contaminated have Nb/Th < 13 and/or Ce/Pb < 20 (and/or MgO < 10 wt.%, not shown). These threshold values are the lower limit of the “mantle composition” defined by the MORB database of Jenner and O'Neill (2012), which is shown as a dashed line and gray field (±1 SD) in both panels. Baffin Island and West Greenland lavas from a metasomatized source (Ba/Th > 100) are marked with a black “X” over the red diamonds. The North Atlantic MORB field is from Gale et al. (2013) and includes only MORB samples from 50 to 80°N that are >500 km from known hotspots (using the hotspot database of King & Adam, 2014).

3.4 Sr-Nd-Hf-Pb Isotopic Compositions

Measured Sr, Nd, and Hf isotopic compositions of the 18 Baffin Island lavas in this data set range from 0.702995 to 0.703946 for 87Sr/86Sr, 0.512920 to 0.513174 for 143Nd/144Nd, and 0.283144 to 0.283287 for 176Hf/177Hf. The ranges for Pb isotopes span 17.5114 to 18.0095, 15.2887 to 15.4291, and 37.455 to 37.971 for 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb, respectively (Table 2).

Figure 7 shows that some Baffin Island-West Greenland lavas with low Nb/Th, Ce/Pb, and MgO also have relatively high 87Sr/86Sr and low 143Nd/144Nd, in some cases approaching radiogenic isotopic values observed in the basement, which has highly geochemically enriched 87Sr/86Sr (0.713758 to 0.823010) and 143Nd/144Nd (0.510737 to 0.511945; Figure 7). Most basement samples extend to lower 206Pb/204Pb than those found in MORB, and Baffin Island-West Greenland lavas with the lowest Nb/Th, Ce/Pb, and MgO also tend to have low 206Pb/204Pb and extend to the unradiogenic values identified in the basement.

Details are in the caption following the image
Sr, Nd, and Pb isotope compositions of Baffin Island (BI) and West Greenland (WG) lavas plotted as a function of the three geochemical indicators for crustal assimilation used here: Nb/Th, MgO, and Ce/Pb. All isotopic data plotted are measured data. Greater degrees of crustal assimilation are associated with lower Nb/Th, Ce/Pb, and MgO; Baffin Island and West Greenland lavas with evidence for crustal contamination also have higher 87Sr/86Sr, lower 143Nd/144Nd, and generally lower 206Pb/204Pb. Lavas (N = 4) identified as the least crustally contaminated using these criteria are marked with a black dot within the red square and outlined with a dashed box. Baffin Island and West Greenland lavas from a metasomatized source (Ba/Th > 100) are marked with a black “X” over the red diamonds. In the bottom panel, five samples (shown with red arrows) with Ba/Th > 100 plot outside the panel. The North Atlantic (50 to 80°N) and global MORB fields are from Gale et al. (2013) and include only MORB samples located >500 km from known hotspots (using the hotspot database of King & Adam, 2014).

After applying the filters for crustal contamination, only four lavas from the Baffin Island-West Greenland suite—AK-8b, DB-13, DB-14, and PI-15—with modern high-precision Sr, Nd, Hf, and Pb isotopic data can be considered “least crustally contaminated” (see Table 2). (We note that an additional five West Greenland lavas fall in this category as well but lack Hf and Pb isotopic compositions determined with modern methods; Larsen & Pedersen, 2009). While it is unfortunate that so few lavas can be considered (near) primary, it is preferable to focus only on those lavas that best reflect the composition of the mantle source.

The four Baffin Island lavas with mantle-like Nb/Th, Ce/Pb, and high-precision Sr, Nd, Hf, and Pb isotopic data, all from this study, plot in the geochemically depleted region of the 143Nd/144Nd versus 87Sr/86Sr and 176Hf/177Hf versus 143Nd/144Nd plots (Figure 8, right panels).

Details are in the caption following the image
Sr, Nd, and Hf isotopic compositions of Baffin Island (BI) lavas from this study (red squares) shown together with previously published data from Baffin Island and West Greenland (WG; both as red diamonds; Jackson et al., 2010; Starkey et al., 2009; Larsen & Pedersen, 2009; Kent et al., 2004, and references therein). Data points shown are the measured isotopic compositions and white and dark gray fields reflect age-corrected and calculated present-day mantle source compositions, respectively (see section S2.3 and Figure S2 in the supporting information). Age correction of the mid-Miocene and modern Iceland lavas is negligible (offset is less than the size of the Baffin Island lava symbols; Figure S3) and the respective fields represent measured data. All isotopic data plotted as symbols here are measured data. Age-corrected data and calculated present-day mantle compositions, are shown as fields in the right-hand side panels. Both crustally contaminated and least crustally contaminated Baffin Island-West Greenland lavas are show in the left-hand side panels, whereas only the least crustally contaminated lavas (Nb/Th > 13, Ce/Pb > 20, MgO wt.% > 10) are shown in the right-hand side panels. Paired Hf and Nd isotopic compositions are available from only two studies—this study (red squares) and Jackson et al. (2010; red diamonds), explaining the smaller data set available for plotting. Mid-Miocene Iceland (darker orange field), modern Iceland (lighter orange field), North Atlantic MORB (50 to 80°N; blue field), and global MORB (light gray field) fields are shown for perspective (Iceland data from GEOROC, http://georoc.mpch-mainz.gwdg.de/georoc/; MORB data from Gale et al., 2013); MORB fields exclude back-arc basin lavas and MORB samples <500 km from nearby hotspots (King & Adam, 2014). Lavas with the highest 3He/4He compositions from Iceland, Galápagos, Hawaii, and Samoa are indicated by the black circles with the letters I, G, H, and S, respectively (see Jackson et al., 2008).

In order to compare the Baffin Island samples with MORB and younger lavas associated with the Iceland plume, we focus on the isotopic compositions calculated for the Baffin Island mantle today (which overlap the measured isotopic ratios), because the age-corrected data are less appropriate for comparison with the significantly younger high-3He/4He lavas from Iceland (all of which are stratigraphically younger than 14.9 Ma; Hardarson et al., 1997; McDougall et al., 1984). In section S2.3 of the supporting information, we provide a method for constraining the present-day composition of the Baffin Island mantle source (i.e., the composition of the source if it had not experienced melt extraction at 60 Ma) to avoid having to compare age-corrected Sr, Nd, Hf, and Pb isotopic compositions in Baffin Island lavas with measured compositions in much younger MORB and Iceland lavas (calculated present-day compositions for Baffin Island lavas are shown in Figures 8, 9, and 12). The four least crustally contaminated Baffin Island lavas with mantle-like Nb/Th and Ce/Pb plot within the field for global MORB located far from hotspots in all radiogenic isotopic spaces (Figures 8 and 9). However, in plots that include 206Pb/204Pb, they are offset from the field for North Atlantic MORB (i.e., MORB samples located between 50 and 80°N that are >500 km away from hotspots), but overlap with it in plots of 143Nd/144Nd versus 87Sr/86Sr and 176Hf/177Hf versus 143Nd/144Nd (Figure 9). The radiogenic isotopic compositions of the four least contaminated Baffin Island lavas do not consistently overlap with the field for mid-Miocene to modern (neovolcanic zone) Iceland lavas, but partially overlap with the geochemically depleted (Sr, Nd, and Hf) and unradiogenic (Pb) portion of the modern (neovolcanic zone) Iceland field. Additionally, they fall on or close to the 4.5 Ga geochron (Figure 9), an observation consistent with that made by Jackson et al. (2010).

Details are in the caption following the image
Sr, Nd, and Pb isotopic compositions of Baffin Island (BI) lavas from this study (red squares) shown together with previously published data from Baffin Island and West Greenland (WG; both red diamonds; Jackson et al., 2010; Starkey et al., 2009; Larsen & Pedersen, 2009; Kent et al., 2004, and references therein). Data points shown are the measured isotopic compositions, while white and dark gray fields reflect age-corrected and calculated present-day mantle source compositions, respectively (see section S2.3 and Figure S2 in the supporting information). Age correction of the mid-Miocene and modern Iceland lavas is negligible (offset is less than the size of the Baffin Island lava symbols; Figure S3) and the respective fields represent measured isotopic ratios. Crustally contaminated and least crustally contaminated Baffin Island-West Greenland lavas are show in the left-hand side panels, whereas only the least crustally contaminated lavas (Nb/Th > 13, Ce/Pb > 20, MgO wt.% > 10) are shown in the right-hand side panels. Mid-Miocene Iceland (darker orange field), modern Iceland (lighter orange field), and North Atlantic MORB (50 to 80°N) (blue field) and global MORB (light gray field) fields are shown for perspective (Iceland data from GEOROC, http://georoc.mpch-mainz.gwdg.de/georoc/; MORB data from Gale et al., 2013); MORB fields exclude back-arc basin lavas and MORB samples <500 km from nearby hotspots (King & Adam, 2014). For all plots that include Pb isotopes, fields for mid-Miocene and modern Iceland are defined using high-precision MC-ICP-MS data only, while MORB fields also include unspiked Pb isotopic data acquired by TIMS. For Pb isotopic data obtained on Baffin Island and West Greenland, both MC-ICP-MS and unspiked TIMS Pb isotopic data are included in the “global plots” (i.e., left-hand side panels), whereas only samples with MC-ICP-MS Pb isotopic data are shown in the right-hand side panels. In the Sr-Pb panel, mid-Miocene Iceland has a narrower range than other panels because the highest and lowest 206Pb/204Pb samples lack Sr isotopic analyses. Lavas with the highest 3He/4He compositions from Iceland, Galápagos, Hawaii, and Samoa are indicated by the black circles with the letters I, G, H, and S, respectively (see Jackson et al., 2008).

3.5 Oxygen Isotopic Compositions

In Figure 10, the oxygen isotopic compositions measured on Baffin Island olivines from this study are shown together with previously published olivine oxygen isotopic data (Kent et al., 2004). The oxygen isotopic data are compared with olivine forsterite content and basalt Nb/Th. The four least crustally contaminated Baffin Island lavas have olivine δ18O indistinguishable from MORB olivines (5.0–5.2 ‰; Eiler, 2001), and olivines from all lavas in this study fall within the range defined by mantle olivine δ18O from Mattey et al. (1994) (5.18 ± 0.28 ‰). However, at low olivine forsterite and low basalt Nb/Th (associated with crustal assimilation), olivine δ18O values in some of the Baffin Island lavas plot outside of the window defined by MORB olivines.

Details are in the caption following the image
δ18O compositions of Baffin Island olivines from this study (red squares) and Kent et al. (2004; red circles) compared with olivine forsterite and Nb/Th. The range of δ18O in MORB olivine (light gray bar) is from Eiler (2001). The range of Nb/Th in MORB is from Jenner and O'Neill (2012) and includes 1σ variation (dark gray bar). Low Nb/Th, which is associated with higher degrees of crustal assimilation, may relate to somewhat higher δ18O. The four Baffin Island lavas that are “least crustally contaminated” (based on having high mantle-like Nb/Th, Ce/Pb, and MgO) also have MORB-like δ18O.

3.6 Helium Concentrations and Isotopic Compositions

Figure 11 summarizes the helium results for olivine in vacuo crushing determinations for Baffin Island-West Greenland lavas from this and previous studies (Graham et al., 1998; Jackson et al., 2010; Stuart et al., 2003; Starkey et al., 2009 and Rizo et al., 2016). Olivine crushing in vacuo is the most common helium extraction method because it primarily releases gas from fluid and melt inclusions, which is the best determination of magmatic helium isotopic compositions, due to the possible presence of cosmogenic and/or radiogenic helium in the olivine matrix. Figure 11 demonstrates that helium concentrations are highly variable in the Baffin Island olivines (a factor of 370), most likely reflecting variable abundances of trapped melt and fluid inclusions in the olivines. The olivine crush experiments for the samples in this study yield 3He/4He ranging from 0.17 to 56.6 RA, encompassing known values for terrestrial, mantle-derived rocks. In general, Baffin Island-West Greenland olivine samples with low 4He concentrations (< 1.0 × 10−9 4He cc STP/g) have lower 3He/4He, due to greater potential for atmospheric contamination in low-4He samples, and greater sensitivity to reduction in 3He/4He by posteruptive radiogenic ingrowth of 4He (Hilton et al., 1995). Two samples—AK-9 and DB-17—plot above the trend defined by Baffin Island-West Greenland lavas in 3He/4He versus 4He space, and, given their low 4He concentrations, were selected for fusion experiments (together with AK-8b) on crushed powders to test for cosmogenic 3He influence.

Details are in the caption following the image
Helium isotopic compositions compared to 4He concentrations for Baffin Island and West Greenland magmatic olivines. Samples with lower helium concentrations tend to have lower 3He/4He, possibly due to greater sensitivity to posteruptive radiogenic ingrowth of 4He. The dashed lines connect the olivine crush experiment data to the respective fusion results for three different samples. The solid line connects a crush experiment on a single olivine megacryst (denoted by an “M” in the symbol) to the crush experiment for muliple smaller olivine phenocrysts from the same lavas (AK-13). In the key, CC signifies crustal contamination.

Basaltic olivines with cosmogenic helium typically yield magmatic helium via crushing and extremely high 3He/4He from fusion, reflecting spallation 3He in the solid olivine (e.g., Kurz, 1986). Sample AK-9, which has the highest 3He/4He crush experiment in this study (AK-9, 56.6 ± 1.1 RA), yielded a 3He/4He of 36.3 RA and 4He concentration of 5.2 × 10−9 cc STP/g by fusion of the powder remaining after crushing. The 3He/4He from the crush experiment of AK-9 is treated with caution due to the low 4He concentration (1.4 × 10−10 cc STP/g) and high (43%) contribution from blank. Another sample with coupled crush/fusion measurements, DB-17, has a crushed 3He/4He of 31.2 ± 0.9 RA (4He = 7.5 × 10−11 cc STP/g) and a fusion 3He/4He of 6.4 RA (4He = 4.7 × 10−9 cc STP/g). Critically, the high 3He/4He value (39.9 ± 0.5 RA) for an olivine crush experiment, determined in sample AK-8b, has the highest 4He concentration (2.3 × 10−8 cc STP/g) and plots within the field of data populated by previously published high-3He/4He lavas in the 3He/4He versus 4He (cc STP/g) plot and is considered the most robust high-3He/4He measurement in this study. A fusion experiment on the AK-8b crushed olivine powder yielded 3He/4He of 20.8 RA (and 4He = 1.84 × 10−8 cc STP/g). In all three samples with paired crushed-powder fusions, the fusion measurements yielded lower 3He/4He than crushing, suggesting that radiogenic helium is a significant contribution. These data demonstrate that cosmogenic helium does not dominate in the olivines and is not a likely contributor to the crushing experiments, because if that were the case one would expect cosmogenic helium to have higher 3He/4He. The lack of high 3He/4He in the fusion measurements does not exclude the possibility of small amounts of cosmogenic helium, but strongly suggests it is not a contribution to the crushing measurements. Olivine typically has extremely low Th and U abundances (ppb), but radiogenic helium can be implanted into the olivine crystal surfaces from the solid matrix (which has ppm levels of Th and U), which is released by fusion and not by crushing (e.g., Jackson et al., 2010; Moreira et al., 2012). The comparison between a megacryst and smaller grain size olivines from the same sample (AK-13) supports the importance of 4He implantation from the groundmass, that is, with greater effect on smaller crystals with fewer melt inclusions.

When focusing only on Baffin Island samples with mantle-like Nb/Th and Ce/Pb (i.e., least crustally contaminated), paired 3He/4He and Sr-Nd-Pb isotopic measurements show that the least contaminated Baffin Island lavas have a distinct radiogenic isotopic composition from the highest observed 3He/4He lavas from Iceland, Galápagos, Hawaii, and Samoa (marked as “I,” “G,” “H,” and “S” in Figures 8 and 9). For example, while the upper envelope of 3He/4He in Icelandic lavas increases with increasing 87Sr/86Sr—where the highest 3He/4He of 37.7 RA is at 0.703465—the measured 87Sr/86Sr of the least contaminated Baffin Island lavas defines a narrow range of lower values (0.703008–0.703021) at all 3He/4He values (0.703009 at 39.9 RA; Figure 12). Thus, the Iceland data form a trend that diverges away from the Baffin Island lavas, and this observation holds for both the measured 87Sr/86Sr ratio and the calculated present-day 87Sr/86Sr for the Baffin Island mantle source. Similarly, a plot of 3He/4He versus 143Nd/144Nd shows that the highest 3He/4He Iceland lavas have lower 143Nd/144Nd (0.512969) than the measured 143Nd/144Nd (0.513128) and calculated present-day mantle ratio, of the least contaminated high-3He/4He Baffin Island lava, sample AK-8b. Finally, paired 3He/4He and 206Pb/204Pb compositions of Baffin Island lavas with mantle-like Nb/Th and Ce/Pb do not overlap with Iceland lavas. The highest 3He/4He Baffin Island lava has Pb isotopic compositions (e.g., measured 206Pb/204Pb = 17.7560) that are less radiogenic than those of the highest 3He/4He Iceland lava (e.g., 206Pb/204Pb = 18.653), an observation that holds for both the measured 206Pb/204Pb and the calculated present-day 206Pb/204Pb for the Baffin Island mantle source. There is no evidence that the least contaminated Baffin Island lavas and Iceland high-3He/4He lavas converge at a common Sr, Nd, and Pb isotopic composition, even if existing data trends are extrapolated to higher 3He/4He. Unfortunately, there are insufficient existing samples with paired 3He/4He and 176Hf/177Hf to make this comparison.

Details are in the caption following the image
Helium isotopic compositions for several hotspots shown as a function of whole rock 87Sr/86Sr, 143Nd/144Nd, and 206Pb/204Pb. Data points shown are the measured isotopic compositions, and white and dark gray fields reflect age-corrected and calculated present-day mantle source compositions, respectively; 3He/4He data are not age corrected. Lavas with the highest 3He/4He in Iceland (yellow field and symbols; note that the yellow field includes all lavas from mainland Iceland including mid-Miocene and Neovolcanic zone lavas) and the least crustally contaminated Baffin Island lavas (red squares) exhibit different Sr, Nd, and Pb isotopic compositions (the comparisons rely on measured isotopic data [red squares] and calculated present-day isotopic compositions of the mantle source of the Baffin Island lavas; see section S2.3 in the supporting information). The least crustally contaminated lavas from Baffin Island have lower 87Sr/86Sr and 206Pb/204Pb, and higher 143Nd/144Nd, than the highest 3He/4He Iceland lavas, suggesting a different high-3He/4He source (see insets). The gray dashed line contains the field for Baffin Island and West Greenland lavas that are crustally contaminated (Nb/Th < 13, Ce/Pb < 20, and/or MgO < 10 wt.%), or are insufficiently characterized to identify potential crustal contamination (e.g., many Baffin Island lavas with 3He/4He data lack Pb concentration [and Pb isotopic] data; Stuart et al., 2003; Starkey et al., 2009). A global data set for oceanic lavas, including MORB and samples from the four hotspots with 3He/4He > 30 RA, are provided for context (fields are adapted from Jackson, Hart, et al., 2007; Jackson et al., 2008).

4 Discussion

4.1 Two Geochemically Distinct High-3He/4He Components in the Iceland Plume or Crustal Assimilation in Baffin Island High-3He/4He Lavas

The highest 3He/4He lavas from Iceland (up to 47.5 RA; Harðardóttir et al., 2018; or 37.7 RA when only considering lavas that have been characterized with paired radiogenic isotope analyses; Hilton et al., 1999), Hawaii (35.3 RA; Kurz et al., 1983, 1982; Valbracht et al., 1997), Samoa (33.8 RA; Farley et al., 1992; Jackson, Hart, et al., 2007; Workman et al., 2004), and Galápagos (30.3 RA; Graham et al., 1993; Jackson, 2008; Kurz et al., 2014; Kurz & Geist, 1999) have distinct radiogenic isotopic compositions (see Figures 8, 9, and 12). Here we show that the radiogenic isotopic compositions of the least crustally contaminated high-3He/4He lavas from 60 Ma Baffin Island document a mantle domain that is geochemically distinct from mid-Miocene Iceland lavas with the highest 3He/4He. Thus, we argue for the presence of two geochemically distinct high-3He/4He components within a single mantle plume. However, it is essential to explore whether the difference in radiogenic isotopic compositions between the least crustally contaminated Baffin Island lavas and Iceland high-3He/4He lavas reflects temporal evolution of the high-3He/4He mantle source sampled by the Iceland hotspot or continental crust assimilation by the Baffin Island lavas.

Crustal contamination is recorded in high-3He/4He continental flood basalts associated with the Icelandic plume at Baffin Island, West Greenland, and East Greenland (e.g., Day, 2016; Larsen & Pedersen, 2009; Lightfoot et al., 1997; Peate, 2003; Yaxley et al., 2004). If the Iceland hotspot has a single high-3He/4He component, one hypothesis is that the high-3He/4He mantle component sampled at the Iceland hotspot has a single Sr-Nd-Hf-Pb isotopic composition over time and that the difference in Sr-Nd-Hf-Pb between Iceland and the least contaminated Baffin Island high-3He/4He lavas is due to melts of the latter having assimilated some amount of continental crust. Radiogenic isotopic compositions for basement samples from West Greenland reported by Larsen and Pedersen (2009)—which are inferred to be similar to the basement underlying the Baffin Island picrites (St-Onge et al., 2009)—allow a test of this hypothesis by investigating the influence of crustal contamination on the radiogenic isotopic compositions of Baffin Island lavas.

The four least contaminated Baffin Island lavas have lower 87Sr/86Sr (0.703009) than the highest 3He/4He (37.7 RA) Iceland lava with available 87Sr/86Sr data (0.703465 for sample SEL97; Hilton et al., 1999). The shift to lower 87Sr/86Sr in the least contaminated, highest 3He/4He Baffin Island lava cannot be explained by continental crust assimilation because assimilation of the local Precambrian crust (which has very high 87Sr/86Sr—0.713758 to 0.823010—compared to the least crustally contaminated Baffin Island lavas, 0.703008 to 0.703021) would only serve to increase the Baffin Island 87Sr/86Sr, not decrease it (Figure 8). Therefore, lower 87Sr/86Sr in the least crustally contaminated high-3He/4He Baffin Island lavas relative to high-3He/4He Iceland lavas must relate to differences in their respective mantle source compositions, an observation that holds for both measured and age-corrected 87Sr/86Sr in Baffin Island lavas, as well as calculated present-day 87Sr/86Sr of the Baffin Island mantle source (Figure 12). This argument does not exclude a small contribution of continental crust assimilation in the four least crustally contaminated Baffin Island lavas. Rather, invoking this would only enforce the argument that Baffin Island and Iceland high-3He/4He lavas have distinct 87Sr/86Sr, because any crustal contamination in Baffin Island lavas would be expected to increase the 87Sr/86Sr, suggesting that hypothetical uncontaminated versions of these lavas would have even lower 87Sr/86Sr relative to the high-3He/4He Iceland lavas.

Neodymium isotopic compositions of Baffin Island lavas lead to a similar conclusion. The measured (and calculated Baffin mantle source today) 143Nd/144Nd of the four least crustally contaminated high-3He/4He lavas from Baffin Island have higher 143Nd/144Nd than high-3He/4He Iceland lavas (Figure 12), an observation that also cannot be explained by crustal assimilation because continental crust—which has very low 143Nd/144Nd (0.510737 to 0.511945) compared to the least crustally contaminated Baffin Island lavas (0.513094 to 0.513128)—would lower the 143Nd/144Nd of the Baffin Island lavas (Figure 8). Thus, the observation of a more geochemically depleted high-3He/4He component in the proto-Iceland plume, compared to the mid-Miocene to modern Iceland plume, is consistent for both 143Nd/144Nd and 87Sr/86Sr. Unfortunately, to few Hf isotopic data exist to verify that his also holds true for 176Hf/177Hf.

In 3He/4He versus 206Pb/204Pb isotopic space, there is no overlap in the 206Pb/204Pb compositions of the least contaminated Baffin Island lavas and the Iceland field (Figure 12). It is equally important to evaluate whether the difference in Pb isotopic compositions between Iceland and Baffin Island lavas relates to continental crust assimilation, because Pb isotopes in basalts can be more susceptible to the compositional effects of crustal assimilation than Sr and Nd isotopes. Indeed, Pb is ~60 times more concentrated in the average West Greenland basement (from Larsen & Pedersen, 2009) than in the least contaminated Baffin Island lavas, whereas Sr and Nd are only ~3 and ~5 times, respectively, more concentrated in the former compared to the latter. Therefore, it is crucial to test the hypothesis that Iceland and Baffin Island high-3He/4He mantle sources actually have the same Pb isotopic compositions and that the apparent shift to lower 206Pb/204Pb in Baffin Island lavas is due to assimilation of continental crust with less radiogenic Pb. To test this hypothesis, a mixing model combines basement material from Larsen and Pedersen (2009) with the composition of the highest 3He/4He lava (with measured Sr-Nd-Pb isotopes) from Iceland (Hilton et al., 1999: see section S3.1 and Figure S4 in the supporting information). While crustal assimilation of an Iceland high-3He/4He lava composition can generate the Pb isotopic compositions of the least contaminated Baffin Island lavas (Figure S4), it also generates strong crustal contamination signatures (i.e., low continental-like ratios) in Ce/Pb and Nb/Th that are not seen in the least crustally contaminated Baffin lavas (section S3.1.). In this light, we find that there is no basement composition in this data set that, through crustal assimilation, can explain the Sr-Nd-Pb isotopic shift from the composition of the highest 3He/4He Iceland lava to the least contaminated Baffin Island lavas while also generating the mantle-like Ce/Pb and Nb/Th observed in the same lavas (Figure S4). Furthermore, the observation of mantle-like Ce/Pb, Nb/Th, and δ18O in the four least contaminated Baffin Island lavas suggests that these four Baffin Island lavas have assimilated very little, if any, continental crust. We conclude that the four Baffin Island lavas are likely very close in composition to their original uncontaminated compositions and that, hence, the Pb isotopic composition of their mantle source must be less radiogenic than (and therefore isotopically distinct from) Iceland high-3He/4He lavas.

It is important to acknowledge that our application of strict trace elements filters, applied to avoid Baffin Island lavas that may have experienced crustal assimilation, may have also filtered out lavas with primary “enriched mantle” (EM) signatures, which have been suggested to exist in the Baffin Island mantle (Kent et al., 2004; Robillard et al., 1992; Starkey et al., 2009, 2012). For example, the enriched mantle (with higher 87Sr/86Sr and lower 143Nd/144Nd) has low Ce/Pb and Nb/Th due to continental crust recycling (Hofmann, 1997; Jackson, Hart, et al., 2007) and any uncontaminated Baffin Island primary melts sampling EM domains with low mantle-derived Ce/Pb and Nb/Th could potentially be eliminated from consideration due to the strict crustal assimilation filters applied to the data set. However, even if enriched mantle lavas with low Ce/Pb and Nb/Th have been filtered out from the Baffin Island data set, this does not negate the finding of a high-3He/4He component in Baffin Island that is more geochemically depleted than the Iceland high-3He/4He component (Figure 12): the observation remains that there are high-3He/4He Baffin Island lavas with lower 87Sr/86Sr and higher 143Nd/144Nd that are more geochemically depleted than high-3He/4He Iceland lavas, and these isotopic differences cannot be explained by crustal assimilation. This supports prior suggestions of a depleted component in the Baffin Island flood basalt suite (Starkey et al., 2009; Kent et al., 2004).

4.2 A Heterogeneous High-3He/4He Component: Implications for a Common Component in the Mantle and the Origins of Its Geochemically Depleted Nature

It is intriguing that, among the highest 3He/4He hotspot localities with 3He/4He > 30 RA (Hawaii, Galápagos, Samoa, and Iceland), the least crustally contaminated Baffin Island lavas have the lowest 206Pb/204Pb and the most geochemically depleted 87Sr/86Sr and 143Nd/144Nd (see Figures 8, 9, and 12). These observations provide important clues to the origin of the high-3He/4He mantle domain. For example, it could be that the extent of geochemical depletion relates to the process that generated high-3He/(U + Th)—and thus enabling preservation of high 3He/4He—in the Baffin Island mantle source. If He is less incompatible than U and Th during mantle melting (e.g., Parman et al., 2005), then greater geochemical depletion will result in higher 3He/4He and higher 143Nd/144Nd, consistent with the highest 3He/4He preserved in Baffin Island lavas with higher 143Nd/144Nd than observed at other high-3He/4He hotspots. However, other studies examining the partitioning of helium during mantle melting suggest that He is more incompatible than U and Th (Heber et al., 2007; Jackson et al., 2013). Thus, an alternative model for the highly geochemically depleted Sr and Nd (and unradiogenic Pb) isotopes in Baffin Island lavas, compared to other high-3He/4He OIB, is that variable quantities of enriched lithospheric material (e.g., recycled oceanic and/or continental crust) were added to an initially homogeneous, geochemically depleted high-3He/4He mantle source (like that seen in Baffin Island lavas) to produce the Sr-Nd-Hf-Pb isotopic variability observed in high-3He/4He (>30 RA) lavas at Hawaii, Galápagos, Samoa, and Iceland (Garapić et al., 2015). A similar conclusion was drawn by Trela et al. (2015) to explain secular cooling of the Galápagos plume as addition of recycled material would decrease the buoyancy (and therefore ascent rate) of rising plume material. Consistent with this alternative scenario, radiogenic 206Pb/204Pb in the high-3He/4He Iceland component, compared to Baffin Island lavas, could result from the addition of a high-U/Pb component. Recycled oceanic crust is an obvious candidate for the high-U/Pb material and would help explain the elevated Ti in high-3He/4He OIB lavas relative to Baffin Island (Garapić et al., 2015) and the recycled atmospheric heavy noble gas signatures in a moderately high-3He/4He Icelandic lava (Mukhopadhyay, 2012) and in high-3He/4He Samoan plume-related lavas in the Lau Basin (Pető et al., 2013). This is further supported by combined trace element modeling and geophysical observations that show a recycled component in mainland Iceland lavas (Shorttle et al., 2014). In this scenario, Baffin Island lavas sample the most pristine (or “least modified”; White, 2015) surviving relic of an early formed, geochemically depleted high-3He/4He mantle domain that has experienced the least overprinting by recycled material over geologic time. Recent measurements of δD in high-3He/4He lavas show that some high-3He/4He lavas retain primordial δD, while others sample a recycled water component, illustrating that high-3He/4He lavas are known to contain signatures of varying amounts of recycled material (Loewen et al., 2019). Heavy noble gases would provide an ideal test of this hypothesis. 129Xe/130Xe data are available for a moderately high-3He/4He (17.2 RA) lava from the neovolcanic zone of Iceland and a Lau Basin high-3He/4He lava (28.1 Ra; Pető et al., 2013), and they indicate both an early Hadean component in the respective mantle sources and the presence of recycled atmospheric heavy noble gases. Unfortunately, the heavy noble gas compositions of Baffin Island lavas have not yet been analyzed, rendering premature the use of heavy noble gases to evaluate whether the Baffin Island high-3He/4He mantle domain has experienced less overprinting by recycled materials compared to other high-3He/4He hotspot lavas. Conclusions regarding a recycled atmospheric heavy noble gas component in the highest 3He/4He mantle domain (e.g., Mukhopadhyay & Parai, 2019) might benefit from measurements targeting the Baffin Island-West Greenland suite.

The addition of recycled material to depleted mantle, as inferred for the high-3He/4He Baffin Island lavas, has implications for the origin of Sr-Nd-Pb isotopic heterogeneity in high-3He/4He lavas and the “common component” sampled by hotspots. In Sr-Nd-Pb isotopic space, different hotspots form “arrays” that converge on a common region, referred to as FOZO (focus zone; Hart et al., 1992) or C (common; Hanan & Graham, 1996), and hotspot lavas that sample this common composition are suggested to host high 3He/4He (Hart et al., 1992). However, the new data from Baffin Island, combined with previously published data from the Iceland hotspot, do not seem to be consistent with convergence on a common high-3He/4He component in Sr-Nd-Pb isotopic space: arrays formed by Iceland and Baffin Island high-3He/4He lavas diverge with increasing 3He/4He in Figure 12. Rather than a homogeneous high-3He/4He domain sampled by all hotspots, a model of heterogeneity in the high-3He/4He domain, that incorporates the addition of heterogeneous recycled materials to a mantle component similar to Baffin Island, may be more consistent with the observed intrahotspot heterogeneity in the high-3He/4He domain(s) sampled by Iceland, and the interhotspot Sr-Nd-Pb heterogeneity observed in the highest 3He/4He lavas from hotspots globally (Jackson, Kurz, et al., 2007; Figure 12).

The origin of the geochemically depleted radiogenic isotopic signatures of lavas with primitive 3He/4He has eluded explanation since the first geochemical characterization of high-3He/4He lavas (Kurz et al., 1982). With the exception of 3He/4He and 182W (but for 182W see section 4.3), the least crustally contaminated Baffin Island picrites resemble MORB in all radiogenic isotopic spaces explored here (Figures 8 and 9; Ellam & Stuart, 2004), as well as 142Nd/144Nd (de Leeuw et al., 2017), 187Os/188Os (Dale et al., 2009), and stable isotopes (e.g., δ18O, this study, and δ56Fe and δ66Zn, McCoy-West et al., 2018). One hypothesis is that the proto-Iceland plume head incorporated significant upper mantle material, which would have been enhanced by concurrent rifting (e.g., Keen et al., 2012), with the result that the upper mantle dominates the non-noble gas isotopic signatures in erupted lavas; the high-3He/4He signature from the deep mantle source was retained due to higher concentrations of helium in the deep mantle relative to the upper mantle (Stuart et al., 2003, 2000; section S3.2 of the supporting information). In this model, the composition of the deep mantle domain contributing high 3He/4He to the depleted mantle is unknown because it has been almost completely overprinted—for everything except for noble gases and possibly W—by mixing with the depleted upper mantle.

Alternatively, an intrinsic depleted component (distinct from the upper mantle MORB source) may reside in the Iceland plume (e.g., Fitton et al., 2003), and if this component hosts elevated 3He/4He, it provides an alternative explanation for the geochemically depleted nature of high-3He/4He material in the Iceland plume. PREMA (Prevalent Mantle) was suggested to be a geochemically depleted lower mantle component sampled by mantle plumes that overlaps with the radiogenic composition of Icelandic high-3He/4He lavas (Zindler & Hart, 1986). One model proposed for the origin of PREMA is that it is the depleted residue of “significant differentiation of the silicate portion of the Earth [that] occurred contemporaneously with core segregation … and might represent the most primitive remaining mantle, having essentially survived unscathed since the earliest days of Earth history” (Zindler & Hart, 1986). In this model, PREMA in the upper mantle convective regime continued to be depleted by crustal extraction and evolved toward depleted MORB mantle (DMM; Zindler & Hart, 1986). If the least contaminated Baffin Island lavas sample PREMA that was preserved in the lower mantle, the short-lived 142Nd/144Nd system (where 146Sm decays to 142Nd, t1/2 = 103 Ma), sensitive to Hadean silicate differentiation, permits investigation of early differentiation that might have generated the geochemically depleted mantle domain with high 3He/4He. However, no resolvable 142Nd/144Nd anomalies are observed in either Baffin Island (de Leeuw et al., 2017) or Iceland (Andreasen et al., 2008; Debaille et al., 2007). The implications of the lack of observable 142Nd/144Nd anomalies in Iceland hotspot lavas (which contrasts with resolvable 142Nd/144Nd variability at other hotspots; Horan et al., 2018; Peters et al., 2018) are not yet clear, but may still leave open the possibility of Baffin Island lavas sampling the depleted residue of Hadean terrestrial differentiation (see section S3.2.), which in turn would be consistent with the primitive Pb isotopic compositions of the least crustally contaminated Baffin Island lavas (Jackson et al., 2010; this study).

4.3 Location of the High-3He/4He Mantle Domains Sampled by the Iceland Plume

While it is important to define heterogeneity that exists within the highest 3He/4He domain in the mantle, it is also important to constrain where the heterogeneous high-3He/4He domains reside within the mantle. Relationships between 3He/4He and geophysical observations at hotspots can provide a clue regarding the location of these domains in the mantle. The hotspot localities with greater contributions from the FOZO-C components (inferred to have high 3He/4He) have lower seismic shear wave velocity anomalies in the shallow (200 km) upper mantle (Konter & Becker, 2012) and higher buoyancy flux than lower 3He/4He hotspots (Graham, 2002; Jackson et al., 2017; Jellinek & Manga, 2004; Putirka et al., 2007), which is consistent with higher 3He/4He hotspots sampling hotter mantle domains than lower 3He/4He hotspots and MORB (Putirka, 2008; Jackson et al., 2017). Here we examine whether the highest 3He/4He hotspot lavas sampled at Baffin Island-West Greenland also sample a mantle source that was hotter than ambient mantle sampled by MORB (Herzberg & Gazel, 2009; Holm et al., 1993; Putirka et al., 2007; Trela et al., 2017). Using the approach of Herzberg and Asimow (2015), we explore the hypothesis of a hotter-than-ambient-mantle high-3He/4He plume by comparing calculated mantle potential temperatures from (1) the least crustally contaminated Baffin Island-West Greenland compositions and (2) highly magnesian MORB from the Siqueiros transform fault that show a clear olivine liquid line of descent. The calculated mantle potential temperatures for the least contaminated Baffin Island-West Greenland primary melts range from 1510 to 1630 °C. This range of temperatures is consistent with previously calculated mantle potential temperatures for proto-Iceland plume lavas from Larsen and Pedersen (2000; 1520 to 1560 °C), Herzberg and Gazel (2009; 1470 to 1650 °C), Hole and Millett (2016; 1480 to 1550 °C), and Putirka et al. (2018; 1630 ± 65 °C). Critically, the Baffin Island-West Greenland lavas yield higher calculated mantle potential temperatures than the high-MgO Siquieros MORB considered here (1300 to 1410 °C), the Siqueiros MORB data set examined by Putirka et al. (2018; 1420 ± 40 °C), the compiled MORB in Madrigal et al. (2016; 1320 to 1390 °C), and average MORB from Cottrell and Kelley (2011; 1320 ± 39 °C).

Hotter mantle potential temperatures result in higher degrees of melting at Baffin Island and West Greenland (10–30%, based on results from the PRIMELT3 calculations for the Baffin Island and West Greenland in Figure 3) relative to high-MgO Siquieros MORB (5–20%, using PRIMELT3 and lavas in Figure 3). Higher degrees of melting in Baffin Island lavas relative to MORB explain lower primary liquid Na2O compositions in the former compared to the latter (e.g., Klein & Langmuir, 1987; see Figure 3). Similarly, higher mantle potential temperatures will result in greater average melting depths for the Baffin Island-West Greenland lavas compared to MORB, consistent with higher calculated primary melt FeO in the former (i.e., Klein & Langmuir, 1987; see Figure 3). Compared to calculated primary MORB melts, higher temperature of melting in the Baffin Island-West Greenland lavas can also explain higher MgO (owing to higher degrees of melting driving melt closer to olivine compositions), lower SiO2 (owing to reduced silica activity at greater melting depths), lower Al2O3 (due to greater extent of melting in the garnet stability field, thereby leaving Al2O3 retained in the source), and lower CaO (because the clinopyroxene-phase volume increases at higher-pressure melting at the expense of olivine and orthopyroxene; for example, Walter, 1998) (see Figure 3). These findings help to explain the rather large differences in the primary liquid major element compositions between MORB (located far from hotspots) and the least crustally contaminated Baffin Island-West Greenland lavas (Figures 2 and 3), and are consistent with the hypothesis that the high-3He/4He domain is sampled by hot plumes (Jackson et al., 2017; Putirka, 2008).

A remaining question is why high-3He/4He hotspots are hotter than both low-3He/4He hotspots (Jackson et al., 2017) and MORB located far from hotspots. One hypothesis is that primitive domains are preserved in deep, dense mantle reservoirs (Deschamps et al., 2011; Jellinek & Manga, 2004; Samuel & Farnetani, 2003), and only the hottest mantle plumes are sufficiently buoyant to entrain this material from the deep mantle (Jackson et al., 2017). A deep dense domain is ideally suited for preserving primitive geochemical signatures that, like 3He/4He, record the earliest history of the planet despite billions of years of mantle convective mixing. For example, the highest 3He/4He lavas from Iceland, Samoa, and Hawaii exhibit negative 182W anomalies relative to the terrestrial standard (Mundl et al., 2017), and these 182W anomalies date to within ~60 Ma of terrestrial accretion. In order for modern OIB to have high 3He/4He and 182W anomalies, there must be domains capable of preserving ancient geochemical signatures within the Earth; however, the exact locations of these domains remain unknown.

Two large low-shear-velocity provinces (LLSVPs) that are consistently observed in seismic tomography studies of the deepest mantle (Garnero & McNamara, 2008; Lekic et al., 2012; McNamara, 2019) may represent storage sites for less degassed and ancient mantle material, as well as younger subducted oceanic or continental crust (e.g., Li et al., 2014). Garapić et al. (2015) observed that the highest 3He/4He hotspots lie above or near the margins of the LLSVPs and used this observation to suggest that these hotspots sample elevated 3He/4He from the LLSVPs. Based on a geographic relationship between hotspots with elevated 3He/4He and the location of LLSVPs, LLSVPs have been argued to host elevated 3He/4He (Williams et al., 2019). If LLSVPs contain primitive geochemical signatures (Tackley, 2000), as well as pockets of heterogenous recycled materials, then variable mixing of primitive and recycled components within LLSVPs could explain the Sr-Nd-Pb-Hf isotopic differences observed between the least contaminated high-3He/4He lavas from Baffin Island-West Greenland and the highest 3He/4He OIB lavas at Iceland, Hawaii, Samoa, and Galápagos (Garapić et al., 2015). More importantly for this study, the juxtaposition (and possible mixing) of ancient high-3He/4He and recycled domains in the plume source could explain the isotopic heterogeneity in the Iceland plume through time. Moreover, ultralow velocity zones (ULVZ; Garnero et al., 2016; McNamara et al., 2010; Rost et al., 2005), which have even slower shear wave velocity anomalies than LLSVPs, provide a second potential long-term storage site for primitive geochemical signatures (Herzberg et al., 2013; Mundl et al., 2017); the three hotspots observed to have 182W anomalies—Hawaii, Iceland, and Samoa—are all associated with ULVZs (Mundl et al., 2017; Mundl-Petermeier et al., 2019). (Note that 182W data are not yet available for Galápagos lavas.) Alternatively, highly viscous mantle domains could lead to the production of isolated convection cells in the mantle, ranging from ~1,000 to 2,200 km depth, called bridgmanite-enriched ancient mantle structures, or BEAMS (Ballmer et al., 2017). Long-term stability of highly viscous portions of the mantle, like BEAMS, may also serve as a storage site for geochemical domains over billions of years. However, further work is needed to explore why material from BEAMS would be preferentially sampled by only the hottest, most buoyant mantle plumes. A contribution from one or more of these domains to rising plume conduits may explain how some isotopic signatures, such as high 3He/4He and 182W, have escaped homogenization and have been observed in mantle-derived rocks that erupted during the Cenozoic.

The core is an additional possible residence site for both elevated 3He/4He and negative 182W anomalies. The possibility of the core as a source of primitive helium in mantle plumes has been amply explored (e.g., Bouhifd et al., 2013; Hofmann et al., 1986; Jephcoat, 1998; Porcelli & Halliday, 2001; Roth et al., 2019). Tungsten is moderately siderophile and therefore partitioned into the core during core formation, resulting in a low Hf/W ratio (and thus negative 182W anomalies) in the core relative to the bulk silicate Earth. Hence, a core contribution to mantle plumes would be observed as negative 182W anomalies in hotspot lavas. Mundl et al. (2017) and Mundl-Petermeier et al. (2019) reported negative 182W anomalies in high-3He/4He Iceland lavas, consistent with a core contribution, whereas Rizo et al. (2016) reported positive 182W anomalies in Baffin Island lavas. However, Kruijer and Kleine (2018) proposed a potential nuclear field shift effect as the origin of the large μ182W found in an Ontong Java Plateau drill core sample (Rizo et al., 2016), leading to these authors speculating about the validity of the large positive μ182W anomaly measured in the Baffin Island sample from the same study. The Rizo et al. (2016) result contrasts with recent results of Mundl-Petermeier et al. (2019), which show slightly negative 182W anomalies in genetically related West Greenland picrites. Given the susceptibility of 182W in primitive Baffin Island and West Greenland lavas (≤62 ppb W; Mundl-Petermeier et al., 2019; Rizo et al., 2016) to being overprinted by continental crust (1,000 ppb W; Rudnick & Gao, 2003), additional 182W analyses from Baffin Island lavas, specifically targeting lavas that are identified as being least crustally contaminated, will be critical for evaluating the presence of μ182W anomalies in the mantle source of Baffin Island lavas.

If additional targeting of the least crustally contaminated Baffin Island lavas reveals anomalous 182W consistent with a core contribution, further investigation of the physical processes and potential geochemical indicators of a core contribution to the mantle will be needed to further assess this hypothesis. For example, it will also require explanation of the lack of extreme highly siderophile element (HSE: Ru, Rh, Pd, Re, Os, Ir, Pt, and Au) enrichment in high-3He/4He lavas expected from a core contribution (e.g., Rizo et al., 2019). It will further be important to understand the mechanism that links anomalous 182W, high 3He/4He, and the hottest/most buoyant plumes (i.e., if the high-3He/4He mantle domain is denser and has anomalous 182W, what is the mechanism responsible for the elevated density and how did it acquire anomalous 182W?).

5 Conclusions

The Iceland hotspot has erupted high-3He/4He for over 60 Myr, providing a natural laboratory for investigation of time-integrated chemical evolution in a high-3He/4He mantle plume. After filtering out Baffin Island-West Greenland lavas that are influenced by continental crust contamination, the least crustally contaminated Baffin Island-West Greenland lavas host a geochemically depleted high-3He/4He component that is more depleted than any other high-3He/4He lavas globally, including high-3He/4He lavas from Iceland. Compositional differences between the least crustally contaminated, high-3He/4He Baffin Island-West Greenland lavas and high-3He/4He mainland Iceland lavas cannot be explained by crustal contamination of the former, indicating temporal evolution of the radiogenic isotopic composition of the high-3He/4He component in the Iceland hotspot. Furthermore, there is no evidence for compositional convergence of Baffin Island-West Greenland high-3He/4He lavas and Iceland high-3He/4He lavas. Therefore, high-3He/4He lavas from the Iceland hotspot do not support a homogeneous high-3He/4He component in the modern mantle. Geochemically distinct high-3He/4He domains within the Iceland hotspot suggests the plume has sampled at least two high-3He/4He domains with distinct Sr-Nd-Pb and, by extension, likely also Hf, isotopic compositions over time. The origin of the geochemically highly depleted radiogenic isotopic compositions in Baffin Island-West Greenland high-3He/4He lavas remains an important outstanding question, but may relate to incorporation of depleted upper mantle during melting in a rift environment and preservation of elevated 3He/4He due to much higher helium concentrations in the high-3He/4He plume compared to the upper mantle. Alternatively, the geochemically depleted nature of high-3He/4He Baffin Island lavas, the highest on record, may reflect a depleted deep mantle domain to which subsequent variable addition of recycled materials has generated the isotopic heterogeneity observed in high-3He/4He lavas from other hotspots. Finally, it is also found that Baffin Island and related West Greenland lavas, which host elevated 3He/4He, record hotter temperatures (1510 to 1630 °C) than Siqueiros MORB (1320 to 1480 °C), consistent with a deep, dense origin for the high-3He/4He mantle domain sampled by the Iceland plume.

Acknowledgments

We acknowledge support from NSF EAR-1624840 (to M.G.J.), NSF EAR-1900652 (to M.G.J.), and NSF OCE-1259218 (to M.D.K). We thank Don Francis for generously providing us access to his collection of Baffin Island lavas. We appreciate helpful discussion and feedback from Roberta Rudnick, Matthew Rioux, Douglas Wilson, and Keith Putirka. Jonathan Pinko is thanked for his help with sample preparation. Rick Carlson's continued generosity is gratefully acknowledged, especially discussions regarding 142Nd/144Nd evolution in the Earth. We acknowledge Al Hofmann for suggesting the use of Nb/Th, instead of Nb/U, in older rocks.  We are grateful for helpful discussion with Maud Boyet while in Paris celebrating one of the author's birthdays. We thank Lotte Larsen and Asger Pedersen for advice and discussion regarding West Greenland samples. We thank C. Herzberg and G. Fitton for thorough and helpful reviews, which greatly improved this manuscript. All data published in this manuscript are available in the EarthChem data repository (https://doi.org/10.1594/IEDA/111373).