1. Recent development of High Pressure Experiments on Composition, physical properties, and thermal state of the core Eiji Ohtani, Graduate school of Science, Tohoku University, Sendai 980-8578, Japan 1

2. Recent development of High Pressure Experiments on the Core 1.Outer and Inner core: less denser than pure Fe: Possible light elements of the core 2.High pressure experiments: high pressure apparatus: 2.1. Phase relations and melting at high pressure: Fe-light element system 2.2. Compression curves and equation of state of metallic iron compounds,   Sound velocity measurements, V p, Vs 3. Current understanding of the core: Some successful model for seismic observations and current limitations of the experimental works : O, Si, and S; H in the core, Carbon in the core

3. 2 Composition? States? Temperature? Center of the Earth 365 GPa 6000~8000 K Central part of the Earth CMB 135GPa, 2000-3500K ICB 330GPa, 5000~7000K V p, V s (km/sec),  (g/cm 3 )

4. Introductions 3.4-5.2 % 2.5-4.2 % Melting temperature of Fe@ICB (Anzellini et al., 2013) 2 Light elements in the Core

5. Poirier (1994) and also see newer version by Hirose et al. (2013)

6. Multianvil Press for High Pressure experiments 37

7. Experimental procedure ・ Multi-anvil high pressure system (3000t, 1500t, 1000t, 700t press) ・ Laser heated Diamond Anvil Cell (LH-DAC) Nd: YAG laser, CO 2 laser, and YLF laser and Temperature measurement by Spectrometer –radiometric method 23

8. Two facilities for SR in Japan Photon Factory フォトン・ファクトリー SPring-8 スプリング８ 19

9. Experimental Method High pressure apparatus :Diamond anvil cell Pressure measurement : Ruby Pressure Scale or Diamond Raman peak shifts (Kawamura et al) Temperature measurement : Spectrometric method using radiation High temperature apparatus : Nd:YLF laser (λ=1.056 μm, heating spot size: 20 μm ） Starting material : Fe-S, Fe-O, Fe-Si (RAREMETALLIC CO.LTD.) Culet size 100, 150, 300 μm Diamond Gasket Re Sample (Fe-4 wt.% Si) Pressure medium (NaCl) Nd:YLF laser (λ=1.056 μm, heating spot size: 20 μm ） 24

10. JEOL JEM-9320FIB Accelerating voltage ： 5-30 kV Single beam A FIB system installed at our lab. in 2005 Ga ion beam gun PC Sample holder Carbon deposition unit SE detector IP Glass-needle preparation N 2 gas

11. Sample room Heated area 50 µm Gasket How we prepare a TEM foil recovered from a DAC experiment?

13. FEI: Versa 3D (FIB) Dual beam (SEM/SIM, 30 kV) Sample pickup: EasyLift Major FIB, SEM and TEM/STEM facilities for common use in Tohoku Univ. FEI: Quanta 3D (FIB) Dual beam (SEM/SIM, 30 kV) Sample pickup: Omniprove JEOL: ARM-200F (TEM/STEM) Cold-FEG (200 kV) Cs-corrector, EDS, EELS FEI: Titan 80-300 (TEM/STEM) FEG (200 kV) Cs-corrector, EDS, EELS JEOL: JEM-2100F (TEM/STEM) FEG (200 kV) EDS JEOL: JSM-7100F (SEM) FEG (30 kV) EDS, EBSD

14. Recent development of High Pressure Experiments on the Core 1.Outer and Inner core: less denser than pure Fe: Possible light elements of the core 2.High pressure experiments: high pressure apparatus: Static vs Dynamic: 2.1. Phase relations and melting at high pressure: Fe-light element system 2.2. Compression curves and equation of state of metallic iron compounds,   Sound velocity measurements, V p, Vs 3. Current understanding of the core: Some successful model for seismic observations and current limitations of the experimental works : O, Si, and S; H in the core, Carbon in the core

15. Melting of Fe-Si-S-O compounds

16. Melting of Fe-Fe 3 S system at 182 GPa Melting After quenching Before melting After quenching Melting Before melting 189.8GPa 170.9GPa 181.8GPa Kamada et al. (EPSL, 2012) 15

17. Melting of Fe-Fe 3 S system up to 182 GPa Kamada et al. (2012) 4

18. Melting relation of Fe-Fe 3 S system at 123 GPa 11 Kamada et al. (2012)

19. 0 10 20 30 Eutectic, at.% Sulfur Morard et al. (2008) 19.9 at% 150 100 50 0 Kamada (2012) 123 GPa, 16(4) at.% Li et al. (2001) 23 at.% Stewart et al. (2007) 19.2 at.% Kamada et al. (2010), 85.6 GPa, 19.6 at.% Pressure, GPa Outer core Campbel et al. (2007) 60 Gpa, 23.5 at.% Figure 8 2550 K 12

20. Asanuma et al. (2010) Melting temperature of FeSi alloy is comparable to that of pure Fe Melting of FeSi alloy up to the Core-mantle boundary pressure 5

21. Phase diagram of the Fe 75 O 5 S 20 alloy Fe-Fe 3 S-FeO solidus temp. is close to Fe-Fe 3 S eutectic temp. Terasaki et al. (2011) 6 K12 Fe-S

22. Phase relation of the Fe-Si-S system Pressure, GPa Temperature, K (Sakairi et al., 2015) 4

23. 0 10 20 30 40 50 60 Pressure, GPa 3000 2500 2000 1500 1000 Temperature, K Magma ocean of the Earth Mercury’s core Martian core A13 A10 K12 T10(S) T10(L) Sol Liq F10 Z&H94 C08 Fig. The melting curves of Fe and Fe-light element systems. The melting curves of Fe-light elements system and solidus of peridotite are shown in this figure. Peridotite: Fiquet et al., 2010 labeled as F10 and Zhang and Herzberg, 1994 labeled as Z&H94). Fe-Si-S (S) Peridotite Fe Fe-S-O (L) Fe-Si-S (L) Fe-S-O (S) Fe-S(S) 6

24. The temperatures at ICB and CMB estimated based on the Fe 75 O 5 S 20 alloy by Terasaki et al. (2011 ）. Pressure, GPa 100 150 200 250 300 350 6000 5000 4000 3000 Temperature, K Solidus T s ad T L ad Liquidus hcp-Fe (Ma et al., 2004) CMB ICB ・ Fe-S-O T(ICB)=4400~5600 K, T(CMB)=3400~4300 K 20

25. Recent development of High Pressure Experiments on the Core 1.Outer and Inner core: less denser than pure Fe: Possible light elements of the core 2.High pressure experiments: high pressure apparatus: Static vs Dynamic: 2.1. Phase relations and melting at high pressure: Fe-light element system 2.2. Compression curves and equation of state of metallic iron compounds,   Sound velocity measurements, V p, Vs 3. Current understanding of the core: Some successful model for seismic observations and current limitations of the experimental works : O, Si, and S; H in the core, Carbon in the core

26. Compression of Fe-Ni-Si-S compounds

27. λ=0.4108(1) Å λ=0.4109(1) Å Typical X-ray diffraction profile for Fe-9.8 wt% Ni-4.0 wt% Si under high pressures at ambient temperature 25 Diffraction patterns of Fe-9.8wt%Ni-4wt%Si at 300 K The hcp phase is stable at the pressure of the center of the Earth 300K A diamond anvil compressed at 374 (407) GPa and 300 K Asanuma et al. (2009, AIRAPT)

28. Fe 0.88 Ni 0.09 S 0.03 Compression curve of Fe 0.88 Ni 0.09 S 0.03 to 335 GPa Sakai et al. JGR (2012) 38

29. Static compression of Fe 0.83 Ni 0.09 Si 0.08 : 407 GPa in Holmes scale in new scale (374 GPa) Asanuma et al. (EPSL, 2011) 37

30. Current pressure generation: Higher pressure Dubrovinsky et al. (BGI/ESRF) (2013; 2015) ~ 700 GPa Sakai et al. (2015) (GRC/SP8) ~700 GPa Fei / Goncharov (GL) ?

31. 7 O Si Takafuji et al. (2004) Sakai et al. (2006) Fe melt coexisting with Pv and PPv Effect of Pressure on Solubility of O and Si in Fe melt coexisting with Pv and PPv 3000K Reactions between molten iron and silicate perovskite and /or post-perovskite Significant amount of Si and O can dissolve into molten iron.

32. Fe melt and Pv Temp vs fO 2 Kawazoe and Ohtani (2005) Fe melt and Pv Temp vs fO 2 Si O Reactions between molten iron and silicate perovskite and /or post-perovskite Significant amount of Si and O can dissolve into molten iron. 27 GPa 8

33. Outer core Inner core Inner core boundary Growth of the inner core 10 Distribution of Si, S, and O between Inner and Outer cores: Partitioning between solid and liquid metals Inner Core Fractionation of Si, S, and O

34. fcc hcp Si O S Solid iron Liquid iron Outer core Inner core Partitioning of S, Si, and O between Solid-Fe and Liquid-Fe fcc-Fe, D(Si) <1 D(S) <1 hcp-Fe, D(Si) >1 D(S) <1 log 10 D(M) = A + B/T + C (P/T) 12

35. 2000K 5500K 4000K 2000K 4000K 5500K Silicon Oxygen Sulfur log 10 D(M) log 10 D(Si) = 0.9847 -310.7/T(K) - 13.833(P(GPa)/T(K)) D(Si) = 1.25 at 330GPa, 5500K D(S) = 0.855 - 2208/T (K) - 36.497 (P(GPa)/T(K) D(S) = 0.018 at 330GPa, 5500K log 10 D(O) ~ -0.11967 = constant (assumed) D(O) ~0.76 at 330 GPa, 5500K log 10 D(M) = A + B/T + C (P/T) Extrapolation of the partition coefficient to ICB (Righter et al., 1997) 13

36. Estimation of the compositions of inner and outer cores (Sakairi et al., 2015 under review) Light elements: S, X wt.%; Si, Y wt.%; O, Z wt.%: ICB = 330 GPa, 5500 K 1)Density deficit of the PREM outer core:  (PREM-OC)at ICB 2)Density deficit of the PREM inner core:  (PREM-IC) at ICB EOS of Fe, FeS, FeSi, and FeO ( Sakai et al., 2014; Fischer et al., 2011;2014; Seagle et al., 2006) 3)Partition coefficients of S, Si, O between solid Fe and liquid Fe (This experiment) Inner core (wt%) Sulfur Silicon Oxygen 0.1-0.2 0-4.3 4.3-0 Outer core (wt%) Sulfur Silicon Oxygen 3.8-8.5 0-3.4 6.2-0 S Si O OC IC OC 14

37. Two models for the composition of the core derived from geochemical and cosmochemical analyses (McDonough, 2014) 10 Model I: O-absentModel II: O-bearing

38. Estimation of the compositions of inner and outer cores Light elements: X wt.% FeS, Y wt.% FeSi, Z wt.% FeO : ICB 330 GPa, 5500 K 1) Density deficit of the outer core: Density of the PREM outer core,  (PREM-OC)at ICB  (PREM-OC) = (1-X-Y-Z)*  (Fe) + X*  (FeS)OC + Y*  (FeSi)OC + Z*  (FeO)OC 2)Density deficit of the inner core:  (PREM-IC) = (1-X-Y-Z)*  (Fe) + X*  (FeS)IC + Y*  (FeSi)IC + Z*  (FeO)IC 3)Partition coefficients of S, Si, O between solid Fe and liquid Fe (This experiment) McDonough (2014) S Si O

39. ESRF ・ Fiquet et al. (2001) Fe-Vp ・ Fiquet et al. (2004) Fe-Vp and other materials ・ Antonangelli et al. (2004) : Fe Anisotropy of Vp ・ Badro et al. (2007) FeS, FeSi etc APS ・ Mao, Lin et al. (2011) Fe, FeSi,….. photon phonon Q E2 k2 E1 k1 Energy transfer E1-E2= E Momentum transfer k1-k2 =Q Q E 1-200meV nm -1 Sound velocity Baron @BL35XU, Spring-8 Sound velocity of Earth materials Ultrasonic Brillouin scattering NRIXS (Nuclear Resonance Inelastic X-ray Scattering) IXS (Inelastic X-ray scattering) 42 Sound velocity of Fe-H alloys: Sound velocity and Birch’s law

40. Sound velocity of Iron Alloys using IXS (BL35XU) Simultaneous measurements of sound velocity and density: Flat panel detector (Image plate) 30 hours stable heating at 2300 K and 20 hours heating at 3000 K Fukui et al. (Rev. Sci. Instr., 2013) 6

41. 22 We need to reconsider the analysis because: Base line of hcp-Fe was wrong. Temperature effect cannot be not ignored FeS hcp-Fe PREM Inner Core FeO Room Temperature FeSi FeS 2

42. 22 We need to reconsider the analysis because: Base line of hcp-Fe was wrong. Temperature effect cannot be not ignored

43. Theory (Steinle-Neumann Et al. (2001) PREM inner core Shock Brown and McQween (1986) at 3000-6000 K NRIXS (Mao et al. (2001) at 300K) IXS Fiquet et al. (2001) at 300 K After Fiquet et al. (2004) Discrepancy: IXS and Shock vs NRIXS ? Temperature effect on Birch law ? Fe 4

44. Previous data-set for hcp-Fe was wrong, and our new data are all consistent, i.e., Vp of hcp-Fe is higher than PREM. The temperature effect cannot be ignored. IC Room Temperature Shock experiment Ohtani et al. (2013) Antonangeli and Ohtani (2015) (Along Hugoniot, i.e. high temperature) All @300 K

45. Hcp-Fe at 163 GPa and 3000 K (Highest pressure and temperature for IXS measurements) (Sakamaki e al., 2016 Sci. Adv., in press) 20

46. Fitting our experimental data by a temperature dependent Birch’s law (Sakamaki et al.,Sci. Adv., in press, 2016) V P (ρ, T) = M ρ + B + A (T − T 0 ) (ρ – ρ*). T 0 to be 300K, so M and B are the coefficients of Birch's law at room temperature, while A and ρ* embody the temperature dependence. The cross-over density, ρ*, indicates a maximum upper bound for the validity of this relation. M=1.160±0.025, B=-3.43±0.29, A=7.2±3.6×10 -5 and ρ *=14.2±1.5. 21 Shock compression

47. hcp-Fe at 300 K Fe 3 S at 300 K FeH at 300 K PREM Inner core Effects of S and H on Birch’s Law of hcp-Fe H effet S effect FeH at ICB Fe 3 S at ICB Pure hcp-Fe at 300 K Vp of dhcp-FeH at high pressure and 300 K (Shibazaki et al., EPSL 2012) Vp of Fe3S at high pressure and 300 K (Kamada et al., Am. Min. 2014) 19

48. hcp-Fe at 300 K Fe 3 S at 300 K FeH at 300 K PREM Inner core hcp-Fe0.9Ni0.1 NIS (Lin et al., 2004) H effet S effect Ni effect Effects of S, H, and Ni on Birch’s Law of hcp-Fe 20

49. hcp-Fe at 300 K Fe 3 S at 300 K FeH at 300 K 5500 K 330GPa 300 K 330GPa 5500 K hcp-Fe 0.9 Ni 0.1 NIS (Lin et al., 2004) PREM Inner core Ni H S Combination of light elements and Ni must reduce both Vp and density of hcp-Fe in order to explain PREM inner core 21

50. The extrapolated Vp and density of hcp-Fe at 330 GPa and 5500 K (at ICB): Greater than those of PREM 24 Shock compression

52. McDonough (2014) O absent model Model II O-bearing model Model I S S(IC) O O(IC) Si Si(IC) O(IC) S(IC) O-bearing model O absent model Our model Models explaining density of the outer and inner cores

53. Fe 3 C (B) (Sh) (L) (O) (K) (L) (T) (F) (Sa) I II Vp and density for two models of the inner core Several assumptions: T effects, Vs, compoounds 16 O absent Core O bearing Core

54. Summary: Limitations Lack of sound velocity data causes ambiguity for estimating the inner and outer core structures from mineral physics ・ Vp and Vs to the inner core P and T conditions ： 330 GPa, 5000~6000 K (164 GPa and 3000 K has been achieved): ・ Data for various iron-light element compounds, such as Fe-Si-O- S-C-H alloys and compounds. ・ Seismic anisotropy in the inner ore: Cij, i.e., measurements of single crystals: ambient condition for single crystal of metallic iron. ・ Measurement of the sound velocity and density of the outer core melts: Melt density is also difficult to measure. Only the data at pressures lower than 70 GPa are available at present.