1 Oceanic plateau of the Hawaiian mantle plume head subducted to the uppermost 2 lower mantle (≤ 96 characters) 3 Short title (≤ 40 characters): Hawaiian plume head in the lower mantle 4 5 S. Shawn Wei1*, Peter M. Shearer2, Carolina Lithgow-Bertelloni3, Lars Stixrude3, and Dongdong 6 ian1 7 1 Depart"ent of Earth and Environmental S&iences, Mi&higan State University, MI 48824, US,. 8 2 Ce&il H. and Ida M. Green Institute of Geophysi&s and Planetary Physi&s, S&ripps Institution of 9 /&eanography, Universit( of California, San Diego, La Jolla, CA 92093, US,. 10 3 Depart"ent of Earth, Planetary, and Spa&e S&iences, Universit( of California, Los Angeles, C, 11 90095, US,. 12 *Correspondence to: S. Shawn Wei (swei6"su.edu) 13 14 Abstract (≤ 125 words) 15 he Hawaiian-$"peror sea"ount chain that includes the Hawaiian vol&anoes is created by 16 the Hawaiian mantle plume. Although the mantle plume hypothesis predi&ts an oceani& plateau 17 produced by massive de&ompression melting during the initiation stage of the Hawaiian hotspot, 18 the fate of this plateau is unclear. We dis&overed a mega"eter-s&ale portion of thi&kened oceani& 19 &rust in the uppermost lower mantle west of the Sea of Okhotsk by sta&king seis"i& waveforms 20 of SS pre&ursors. We propose that this thi&9 crust represents a ma:or part of the oceani& plateau 21 that was created by the Hawaiian plume head about 100 Ma ago and subducted 20–30 Ma ago. 22 /ur dis&overy provides te"poral and spatial clues of the earl( history of the Hawaiian plume for 23 future plate re&onstructions. 1 24 25 One sentence summary (≤ 150 characters): 26 he oceani& plateau created by the Hawaiian mantle plume subducted into the Ka"&hatka 27 rench and rea&hed the lower mantle beneath Siberia. 28 29 2 30 While earthquakes and vol&anis" at plate boundaries are well explained with the theory of 31 plate te&toni&s, explaining intra-plate hotspot vol&anoes requires the mantle plume hypothesis 51, 32 27. This hypothesis posits deep-rooted and relativel( fixed plumes of hot material upwelling 33 through the mantle from the deep Earth and a&&ounts for the age-progressive surfa&e expression 34 known as the Hawaiian-$"peror sea"ount chain. As the Pa&i>& Plate moves northwest 53, 47, 35 the newest vol&anoes are found in Hawaii to the southeast, and the oldest sea"ounts are near the 36 <a"&hatka-,leutian trench junction in the northwest. The ~47 Ma bend of the sea"ount chain is 37 usuall( attributed to a change in the Pa&i>& plate motion 557. The history of the Hawaiian- 38 $"peror sea"ount chain is criti&al for understanding Earth’s interior evolution and plate 39 te&toni&s. In the classi&al view, a mantle plume consists of a large head (>2,000 km a&ross) and a 40 thin tail (~200 km wide7 567. The plume head generates a large igneous province (LIP), such as 41 the Ontong-Java oceani& plateau or the De&&an Traps. The plume tail creates an age-progressive 42 intra-plate vol&ani& chain. Several eCorts have been made to associate ancient LIPs to hotspot 43 vol&anoes 577. For instance, the De&&an Traps are considered to result from the head of the 44 Reunion mantle plume surfa&ing more than 68 Ma ago 587. However, the fate of the Hawaiian 45 "antle plume head and resulting oceani& plateau is unknown due to the debatable early history 46 of the Hawaiian-$"peror sea"ount chain. 47 he Hawaiian-$"peror sea"ount chain entered the Ka"&hatka subduction zone based on a 48 variet( of plate re&onstructions 53, 47. One proposal pla&es this event as the cause of the cusp 49 between the Kurile-<a"&hatka and the Aleutian-,laska trenches 597. The subduction of the 50 sea"ounts generates arc lavas with geoche"i&al signatures si"ilar to oceani& island basalts on 51 the Ka"&hatka Peninsula 5107. The oldest surfa&e portion of the Hawaiian-$"peror chain, the 52 Mei:i Guyot (older than 81 Ma) and Detroit Sea"ount (76–81 Ma) 5117 are about to subduct into 3 53 the Ka"&hatka Trench (Digure 1). But whether the older parts of the sea"ount chain, parti&ularl( 54 the plume head, also subducted into the dee! mantle or stayed on Earth’s surfa&e is debated 512- 55 147. The fate of the Hawaiian plume head is criti&al to the origin of the mantle plume, whi&h 56 provides a te"poral constraint on the longevit( and persistence of che"i&al chara&teristi&s of 57 $arth’s deep mantle. Furthermore, the subduction of the expe&ted oceani& plateau caused by the 58 -awaiian plume head ma( have changed plate motions. Niu et al. (127 proposed that the 59 &ollision of this oceani& plateau with the Ka"&hatka Trench was responsible for the Pa&i>& Plate 60 reorientation that resulted in the 47-Ma bend in the Hawaiian-$"peror chain. 61 More i"portantly, the fate of this oceani& plateau is criti&al for understanding the role of 62 oceani& plateaus in building continental lithosphere and in mantle conve&tion. Due to their 63 excess crustal thi&kness and volume, oceani& plateaus are thought to be more diH&ult to subduct 64 than individual sea"ounts 5157. Be&ause the Yakutat terrane southeast of Alaska is the onl( 65 oceani& plateau that is currentl( undergoing subduction 5167, whether oceani& plateaus were 66 &om"onl( subducted in the past is unclear. By anal(Ging ophioliti& basalts in Ka"&hatka, 67 Portnyagin et al. (147 proposed that the Hawaiian plume head, or at least part of it, was a&&reted 68 to the forearc of Ka"&hatka. This me&hanis" provides an i"portant way to grow continental 69 &rust 577. In contrast, a seis"i& study of P-to-S waves converted at seis"i& dis&ontinuities 70 (re&eiver functions) in South A"eri&a suggests that an oceani& plateau with a thi&kness of at 71 least 13–19 km has subducted to ~100-km depth and is responsible for the Pa"pean flat sla8 72 5177. Geodyna"i& models also show oceani& plateaus can subduct into the upper mantle, 73 resulting in slowing down subduction 5187, forming a flat slab 5197, elevating surfa&e topography 74 5207, and generating dyna"i& uplift 5217. Compared to subducting normal oceani& crust with a * 75 thi&kness of 6–7 km, the input of thi&9 oceani& plateaus might also change, at least locally, 76 "antle composition and dyna"i&s. 77 ,lthough mantle plume conduits have been suc&essfull( i"aged using seis"i& tomography 78 with dense datasets 5227, oceani& plateaus potentiall( subducted into the lower mantle have a 20– 79 40 km crustal thi&kness that is s"aller than the resolution in most tomographi& studies. Due to a 80 la&9 of data, the tomography resolution in northeastern Siberia is parti&ularl( low in both global 81 5237 and regional 5247 i"ages. Seis"i& reJe&ted waves are more sensitive to sharp boundaries 82 and provide a more eCe&tive tool to dete&t s"all-s&ale compositional heterogeneities in the dee! 83 "antle. Many seis"i& reJe&tors in the lower mantle have been i"aged globall( and attributed to 84 segments of subducted crust 525-277. But ancient oceani& plateaus have not been dete&ted in the 85 lower mantle, in part due to the li"ited data coverage in regions the( are expe&ted. 86 We sta&ked SS pre&ursors (SdS) from 45 years of global seis"i& data to dete&t seis"i& 87 reJe&tors in the lower mantle 5287. The SdS seis"i& phase is the underside S wave reJe&tion off 88 the d-km dis&ontinuity, whi&h arrives before the surfa&e-reJe&ted SS phase (Digure S1,). 89 Be&ause SS pre&ursors sa"ple the midpoints between earthquakes and seis"i& stations, the( 90 provide good data coverage for re"ote regions and are widely used to i"age seis"i& 91 dis&ontinuities in the upper and mid mantle 5297. Besides the ma:or seis"i& dis&ontinuities 92 extending globally, previous observations dete&ted many s"aller-s&ale reJe&tors using SS or PP 93 pre&ursors 526, 307. 94 We focus on a seis"i& reJe&tor observed at ~810-km depth west of the Sea of Okhotsk, 95 whi&h was previousl( dete&ted by li"ited data of PP pre&ursors 5307. The reJe&tor has a width 96 on the order of 1,000 km and a depth varying from 780 to 820 km a&ross (Digure 2). When 97 &ompared to global tomography models 5237, the 810-km reJe&tor appears to coincide with the 3 98 <a"&hatka slab, whi&h is the ancient Pa&i>& Plate subducted along the Ka"&hatka Trench 99 5Digure 3B). Due to the li"ited resolution of tomography models, determining whether the 100 reJe&tor is above or at the slab surfa&e (top interfa&e7 is challenging. The exa&t shape of this 810- 101 9" reJe&tor is unclear be&ause of the wide Fresnel zone (~1,000 km a&ross) and the low 102 horiGontal resolution of SS pre&ursors.