What is the metallicity of our reference , the ?

Gaël Buldgen

University of Geneva

September 2019

1 Introduction

The Sun as a benchmark star

The role of the Sun: Well-studied, helioseismic constraints, neutrino uxes, testbed for physical ingredients. The Sun is used as a reference: Metallicity scale, Enrichment laws, SSM framework, Paved the way for using solar-like oscillations.

Most of our models will include some ingredients that have been calibrated on the Sun. Thus, if you change the way you model the Sun, you impact stellar physics as a whole.

But how well do we know the Sun?

2 Solar abundances

1D Models of the 20th century

1D solar atmosphere models Theoretical models: MARCS, Kurucz, ... 1 Hydrostatic, 2 MLT type convection, 3 LTE. Usually, semi-empirical models were used e.g. Holweger-Müller model (1974). Led to high abundances of C,N and O (AG89, GN93, GS98).

Pereira et al. (2009) 3 Solar abundances

Advent of 3D Models and abundance revisions

Revision of the abundances: Hydrodynamical model, Non-LTE corrections, improved atomic data, Careful selection of lines, Use of all indicators.

⇒ 30% reduction of Z ! Asplund et al. (2009) 4 Solar abundances

The solar modelling problem

A brief history of Standard Solar Models Before 2004, high metallicity solar models (Z = 0.0182): 1 Correct position of the BCZ, 2 Correct abundance in the CZ, 3 Sound Speed prole relative dierences of up to 0.006. (From Kosovichev & Fedorova 1991, 1993, Vorontsov et al. 1991) But: slow degradatation as physical ingredients were updated. From 2004, downward revision of the solar Z:

1 Wrong position of the BCZ,

2 Wrong Helium abundance in the CZ,

3 Sound Speed prole relative dierences of up to 0.02.

5 Solar abundances

Discussions of the revised abundances

Study by Caau et al. 2011: Higher C,N,O abundances, closer to the “old values”. Subsequent analyses by Scott et al. 2015a, Scott et al. 2015b, Grevesse et al. 2015, Lind et al. 2017, Bergemann et al. 2017, Amarsi and Asplund 2017, Nordlander and Lind 2017, Amarsi et al. 2018, Amarsi et al. 2019: conrm low C,N,O abundances. What causes the discrepancies? 3D Models? No! Stagger and Co5bold agree (Beeck et al. 2012). Selection of lines, blends, molecular lines. If the same inputs are used, the same results are obtained.

What about meteorites? Dierentiation, they are not substitutes to photospheric abundances! (N.Grevesse, private communication)

6 Solar abundances

Neon revision and latest changes

The problem of Neon: No photospheric lines. ⇒ Inferences from coronal lines, solar wind, ... Variations with activity!

Dicult measurements, Ne/H unacessible, Ne/O measured, Antia & Basu (2004): increase of 400% to solve the “solar problem”. Landi & Testa 2015 + Young 2018: increase of Landi & Testa (2015) 40% of Ne/O. 7

The solar modelling problem as seen in seismology:

Current state of the issue, for various abundance and opacity tables...

8 Helioseismology

Inferring Z from helioseismic data I

Constraints from seismic inversions: Inversions can only constrain variables from the acoustic structure : 2  ∂ lnP  ρ, c or Γ1 = for example. ∂ lnρ S However, assuming an E.O.S, one has:

δΓ ∂ lnΓ  δP ∂ lnΓ  δρ ∂ lnΓ  1 = 1 + 1 + 1 δY Γ1 ∂ lnP ρ,Y,Z P ∂ lnρ P,Y,Z ρ ∂Y P,ρ,Z ∂ lnΓ  + 1 δZ, ∂Z P,ρ,Y

thus allowing for inversions of Y, the helium abundance, or Z, the metallicity. Previous studies by Takata & Shibahashi (2001), Antia & Basu (2006) and Vorontsov et al. (2013). 9 Helioseismology

Inferring Z from helioseismic data II

Initial attempts in Takata and Shibahashi (2001) using density and Γ1 kernels.

Impossibility to conclude because of the errors bars. However, the conclusion mentions that the three last point are consistent with a 30% reduction of the solar metallicity. 10 Helioseismology

Inferring Z from helioseismic data III

This inversion (Buldgen et al. 2017c) favours a low metallicity (as in Vorontsov et al. 2013).

0.02 Reference Models Outliers GN93 0.018

0.016 Suboptimal parameters set 0.014 AGSS09 Z 0.012

0.01 Optimal parameters set

0.008

0.006 0 1 2 3 4 5 6 GN93 GN93 GN93 GN93 GN93 OPLIB OPLIB x1.05 OPLIB OPAL OPAL FreeEOS FreeEOS FreeEOS CEFF FreeEOS Diff x0.75. 11 Helioseismology

Potential solutions to the solar modelling problem I (Buldgen et al. 2019, in prep.)

Combination of: Neon increase from Landi & Testa (2015) and Young et al. (2018), extra-mixing and opacity modication (from A. Pradhan)

12 Helioseismology

Impact on stellar physics

A few illustration of the potential impact of the solar problem 1 Solar reference for the chemical abundances, 2 Revision of key ingredients: opacity, EOS, screening factor (Mussack & Däppen 2011, Bailey et al. 2015), 3 Transport of chemicals (see talk by R. Hirschi), 4 Angular momentum transport processes (see talk by P. Eggenberger), 5 Impact on asteroseismic modelling and characterization of stellar populations in the (see talk by C. Chiappini).

The role of calibrator of the Sun implies that changing its ingredients impacts a wide range of stellar models.

13 Conclusions and prospects

Conclusion and perspectives

In conclusion Still a problem: Will new opacity computations do it? Maybe. (Pradhan 2017, Zhao 2017, Pain 2019). What about the BCZ: Extensively studied (see e.g. Hughes 2007 and references therein) Is that it? No: Microscopic diusion, EOS improvements, convection, instabilities, early history (see also Zhang et al. 2019)... What is clear? Stop using GN93 and GS98. (listen to Nicolas Grevesse)

The solar problem is not purely an issue of abundances, rather an issue of other modelling ingredients. No signicant variations found in 2015 by AGSS. Its impact reaches beyond the range of solar models.

14 Conclusions and prospects

Thank you for your attention!

15 Conclusions and prospects

Potential solutions to the solar modelling problem II Other approach: build a seismic model and try to see what its properties may be (Buldgen et al. In prep).

Help li€ some degeneracies and drive revisions of physical ingredients. 16 Conclusions and prospects

Opacity kernels

Based on Tripathy & Christensen- Dalsgaard (1998). Assumes linear behaviour with respect to small κ perturbation. Vinyoles et al. (2017) Allow for a static analysis of the required changes in opacity to match helioseismic constraints. 17 Conclusions and prospects

The current state of the issue

18 Conclusions and prospects

The current state of the issue

19 Conclusions and prospects

The current state of the issue

20 Conclusions and prospects

Combining seismic information I

2 Combining: A, S5/3, c , Y, position of BCZ, mCZ... (Buldgen et al. submitted to A&A)

Similar to Christensen-Dalsgaard & Houdek (2010), Ayukov et al. (2011), Christensen-Dalsgaard et al. (2018). 21 Conclusions and prospects

Combining seismic information II

Combination of: Neon increase from Landi & Testa (2015) and Young et al. (2018), extra-mixing and opacity modication (from A. Pradhan)

22 Conclusions and prospects

Combining seismic information III

It seems: Opacity increase too high and perhaps too steep + wrong ∇T transition in overshooting region (improve on Christensen-Dalsgaard 2011) (Work by Rempel et al. 2004, Zhang et al. 2012) 23 Conclusions and prospects

Inferring Z from helioseismic data - Tests

Tests on articial data (same ν, same σν ) to ensure accuracy.

Reference models 0.022 Inverted results 0.021 Target value 0.02 0.019 0.018 Z 0.017 0.016 0.015 0.014 0.013 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Model Number 24 Conclusions and prospects

Extended solar calibration

Add additional free parameters and constraints to the solar models to expand & the calibration procedure. (Ayukov Baturin 2013, 2017) 25 Conclusions and prospects

Considered opacity modication

0.15

0.1

0.05

0 6.2 6.4 6.6 6.8 7 7.2

26 Conclusions and prospects

Other classical diagnostics

rConv/R YConv Helioseismic measurements 0.713 ± 0.001 0.2485 ± 0.0035 SSM (AGSS09, Free, OPAL) 0.720 0.236 SSM (AGSS09, Free, OPLIB) 0.718 0.230 SSM (AGSS09, Free, OPAS) 0.717 0.232 SSM (GN93, Free, OPAL) 0.711 0.245 SSM (GN93, Free, OPLIB) 0.708 0.240

27 Conclusions and prospects

Standard Models with new opacities - Frequency ratios

0.095 0.16 Obs Obs GN − OP AL 0.09 93 GN93 − OP AL AGSS09 − OP AL AGSS09 − OP AL GN93 − OPLIB 0.15 GN93 − OPLIB 0.085 AGSS − OPLIB 09 AGSS09 − OPLIB r02, r13 ⇒ AGSS09 0.08 0.14 favoured! c2 inversions still 0.075

02 favour GN93. 13 r 0.13 r 0.07 BCZ wrong for both 0.065 0.12 AGSS09 and GN93. 0.06 YS very low for 0.11 AGSS09. 0.055 ⇒ Need new diagnostics. 0.05 0.1 1500 2000 2500 3000 3500 1500 2000 2500 3000 3500 ν (µHz) Hz ν (µ ) 28 Conclusions and prospects

Inversions of the convective parameter for Standard Solar Models 0.1 AGSS09 − OP LIB − ZXGN Base of the solar 0.08 GN93 − OP LIB convective zone

0.06

0.04

0.02 Model

A 0 −

Sun −0.02 A −0.04

−0.06

−0.08

−0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Position r/R

The compensation is related to the heavy-element mixture. 29 Conclusions and prospects

Inversions of the convective parameter for Standard Solar Models

0 ATot AGSS09 −0.2 AT AGSS09 Aµ AGSS09 ATot −0.4 GN93 AT GN93 µ AGN " −0.6 93 r R ! X Y

A −0.8

−1

−1.2

0.5 0.55 0.6 0.65 0.7 Position r/R 30 Conclusions and prospects

Inversions of the convective parameter for Standard Solar Models

0.3 AGSS09 − ATOMIC − DOB Base of the solar GN93 − ATOMIC convective zone 0.25 AGSS09 − ATOMIC GN93 − ATOMIC + DOB 0.2

0.15 Model

A 0.1 − Sun

A 0.05

0

−0.05

−0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Position r/R

The compensation is also related to the temperature gradient. 31 Conclusions and prospects

Relative dierences OPLIB-OPAL

0.15

0.1

0.05 OPAL κ

/ 0 OPAL κ

− −0.05 OPLIB κ −0.1

−0.15

−0.2 3.5 4 4.5 5 5.5 6 6.5 7 7.5 log T

32 Conclusions and prospects

Metallicity Inversions for the Solar Envelope

Metallicity kernels can thus be derived to estimate Z in the envelope.

3 20,0 KZ,A,Y 30,0 2 KZ,A,Y 20,1 KZ,A,Y 25,2 KZ,A,Y 1 n,l Z,A,Y

K 0

−1

−2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Position r/R

33 Conclusions and prospects

Appendices Helioseismology - Hare-and-Hounds exercises

0.01

0

−0.01 Reference

S −0.02 − Reference S T arget S −0.03

−0.04 Real Differences SOLA Results SOLA Results (Less regularized) −0.05 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 r/R 34 Conclusions and prospects

Appendices Helioseismology - Kernel ts

30 AGSS09-OPLIB (SOLA) Averaging Kernel Target Function 0.005 25

0 20

−0.005 15

Base of the solar ) M odel r S convective zone ( − M odel S K

Sun −0.01

S 10

−0.015 5

−0.02 0

−0.025 −5 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 r/R Position r/R 35 Conclusions and prospects

Links with opacity and chemical composition

0.02 Base of the solar 0.015 convective zone

0.01

0.005 GS98

0 M odel

S −0.005 − M odel S Sun S −0.01

−0.015

−0.02 AGSS09 − F ree − OP LIB AGSS − F ree − OP LIB ff 09 + TurbDi AGSS09 AGSS09 − F ree − OP LIB + 10% at BCZ −0.025 GS98 − F ree − OP LIB −0.03 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 r/R Entropy inversions hint directly at inaccuracies in the radiative zone. 36 Conclusions and prospects

Parameters of the solar models with modied opacities and additional mixing used in this study

(r/R)BCZ (m/M)CZ YCZ ZCZ Y0 Z0 Opacity Abundances Diusion 0.7122 0.9757 0.2416 0.01385 0.2692 0.01494 OPAL+Poly AGSS09Ne Thoul 0.7129 0.9761 0.2427 0.01383 0.2678 0.01483 OPAL+Poly AGSS09Ne Paquette 0.7106 0.9762 0.2425 0.01383 0.2685 0.01466 OPAL+Poly AGSS09Ne Thoul+DTurb 0.7106 0.9762 0.2374 0.01359 0.2645 0.01490 OPAS+Poly AGSS09 Thoul+DTurb 0.7121 0.9756 0.2460 0.01376 0.2696 0.01500 OPAL+Poly AGSS09Ne Thoul+DTurb − Prof 0.7118 0.9757 0.2437 0.01381 0.2692 0.01495 OPAL+Poly AGSS09Ne Thoul+Ov − Rad 0.71056 0.9751 0.2438 0.01381 0.2700 0.01506 OPAL+Poly AGSS09Ne Thoul+Ov − Ad

37