Talk contents Part A: Introduction and methods - Caltech
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Talk contents Part A: Introduction and methods 1. Observational motivation 2. Population synthesis principle 3. Input physics: global models 4. Initial conditions 5. Observational biases Part B: Results and perspectives 1. Compute the population: individual systems 2. Overview of statistical results 3. Comparisons with observations 4. Perspectives and conclusions
Compute the synthetic population Andrews+2018 Initial Conditions: Probability Models of individual Global end-to-end form- distributions of disk properties processes ation & evolution model Disk gas mass From Disk dust mass Accretion, migration, … Disk properties planet properties observations Disk lifetime Draw and compute synthetic population New instrumentation better observational constraints Apply observational detection bias Predictions (going back to the full synthetic population) Observed population Stat. Comparison: Observable sub-population Model - Frequencies solution No match: change - Orbits, masses, radii, luminosities - Architecture, multiplicity Match found parameters, improve - Correlations model, reject model ….. One learns a lot even if a synthetic population does not match the observed one!
Global model: Example outcome Numbers are Bern Generation III model -[Fe/H] -Mgaz,0 [MSun] -Msolids,0 [MEarth] Setup/parameters -solar-mass stars -1000 systems (stars) -100 ini6al embryos per system -embryo mass 1 Mluna -uniform in log out to 40 AU at t=0. -viscosity α=0.002 -opacity red. factor 0.003 Gas-dominated giant (Menve/Mcore >=1) (“Jovian”) Protoplanet lost during formation and evolution process (accreted by Volatile rich planet, with H/He (“Neptunian”) another more massive protoplanet, Volatile rich planet, without H/He (“water world”) ejected, collided with the star). Iron/silicate planet with H/He (“H/He terrestrial”) Black bar: peri- to apoastron Iron/silicate planet without H/He (“Earth-like”) distance (showing eccentricity)
A global model in action: low solid mass Initial conditions -initial disk gas mass: 0.017 Msun -initial solid mass: 57 MEarth Class 1 architecture
A global model in action: mid solid mass Initial conditions -initial disk gas mass: 0.042 Msun -initial solid mass: 100.1 MEarth Class 2 architecture
Class 2. The migrated sub-Neptune systems t=5 Gyr Peas-in-a-pod (Weiss+2018, Millholland+2017): Misra+2021
A global model in action: high solid mass Initial conditions -initial disk gas mass: 0.042 Msun -initial solid mass: 245 MEarth Class 3 architecture
A global model in action: very high solid mass Initial conditions -initial disk gas mass: 0.044 Msun -initial solid mass: 327 MEarth Class 4 architecture
Class 4. The dynamically active giants
Manara et al. 2016 2. Overview of statistical results
Formation of the a-M diagram Ini6al condi6ons -solar-mass stars -1000 systems (stars) -100 ini6al embryos per system -ini6al mass 1 Mluna Emsenhuber et al. 2021b (NGPPS II, arXiv: 2007.05562)
Nominal population 1 M☉ Distant super t=5 Gyr Distant Jupiters Hot Cold Jupiters super Jupiters Jupiters Planetary desert Hot Cold Jupiters Close-in Jupiters low-mass ? Planetary desert wet Close-in low-mass ice worlds HZ dry Several fundamental features of the observed popula6on can be recovered. The varia6ons of the disk ini6al condi6ons over a range likely occurring in nature leads to a large diversity of planetary systems, similar as observed.
10 Number of system Number of system 102 Number of syst Number of syst 10 2 Statistical overview 10 for 1 M☉ 102 2 1 Fundamental 10 demographical results 101 10 1 101 • Synthetic planetary system contain on average 0 8 planets more massive than 1 M⊕ 10 including all orbital distances (no obs. constraints yet). 0 5 10 0 10 0 2 4 6 • 10 Fraction of systems 0 with giants 5 planets: 10 18 % (all orbital distances), 0 2 only 4 1.6 %6at >10 AU. Number of planets Number of planets anets • Systems with giantsNumber of average contain on planets 1.6 giant planets.Number of planets 3 Giant t 10 Giant Number of systems 3 10 10 embryos Number of systems 20 10 embryos embryos 2 Initial number of 10 50 20 embryosembryos embryos per disk 2 10 50 embryos 100 embryos 1 100 embryos 10 101 The result is independent of the initial nb of embryos per disk. 0 1 2 3 3 0 1 Number of planets 2 3 Only 5 stars out of 1000 have 3 giant planets. anets Number of planets rent types of planets, for the four multi-embryo populations presented in this study. In each panel, ymofanother to make •di↵erent Low-mass them typesplanets more of planets, visible. We for theMfour (0.3-3 seehabitable for example ⊕) inmulti-embryo that for 44the populations zone: giant %presentedplanets, of stars. outstudy. inMean this ofmultiplicity theIn 1000 each panel, 1.3 (rather planets. About tally from an equal another to makenumber them (100 moreeach) have visible. Weone see or fortwo giants.that example Less for than ten out the giant of theout planets, 1000 of the 1000 ynumber low). Mean per system giant planets. [Fe/H] Aboutthatan of stars occurs. with habitable planets -0.11. Different from Solar equal number (100 each) have one or two giants. Less than ten out of the 1000 System. highest number per system that occurs.
What sets the outcome? 25 The most important ini6al Number of planets in system with M ≥ 1 Me 4 10 condi6on is the mass of 20 solids ini6ally present in the disk. Total system mass [Me] 103 Solar system 15 Grey lines: efficiency of planetary system 10 forma6on(including H/He) 102 per system 10 We need 2-3 MMSN to 1 1 form the Solar System 10 5 mass. 0.1 100 0 1 2 3 10 10 10 Initial mass in solids [Me] Another important ini6al condi6on is external disk photoevapora6on, seRng the disk life6me. It depends on stellar birth environment and Adapted from Mordasini+ in prep. affects the emerging system architecture (e.g., Winter et al. 2020).
Mayor et al. 2011 3. Comparisons with observations
Comparisons with observations Andrews+2018 Initial Conditions: Probability Models of individual Global end-to-end form- distributions of disk properties processes ation & evolution model Disk gas mass From Disk dust mass Accretion, migration, … Disk properties planet properties observations Disk lifetime Draw and compute synthetic population New instrumentation better observational constraints Apply observational detection bias Predictions (going back to the full synthetic population) Observed population Stat. Comparison: Observable sub-population Model - Frequencies solution No match: change - Orbits, masses, radii, luminosities - Architecture, multiplicity Match found parameters, improve - Correlations model, reject model ….. One learns a lot even if a synthetic population does not match the observed one!
Statistical comparison with HARPS survey Observed Nb of planets: 161 Mayor et al. 2011 Nb of stars w planets: 102 Lines: detec+on probabili+es 100% Multiplicity: 1.6 95% Synthetic biased [Earth Mass] 80% 1000. 60% Nb of planets: 317 HARPS: high 40% Nb of stars w planets: 204 accuracy radial 20% Multiplicity: 1.6 velocity planet 10% searcher. 100.0 5% 2% • Agreement: similar global structure: relative distribution (concentrations, voids) M2sini 10.0 • Agreement: Mean multiplicity GTO Survey: 822 system architecture. solar-like stars. • Disagreement: Factor 2 in Known bias. absolute number. Poss. 1.0 explanations: Initial conditions? See also recent 10+0 10+1 10+2 10+3 10+4 Cluster environment (cf. Winter results of California et al. 2020)? Legacy Survey Period [days] • Disagreement: Hot Jupiters. (Rosenthal+2021, Poss. explanation: Kozai plus Fulton+2021) tidal circularisation channel missing in model. Adapted from Emsenhuber, Mordasini, Udry, Mayor, Marmier + in prep. (NGPPS VII)
Quantitative comparison mass distribution 80 Mass distribution detected planets 0.8 80 70 • Agreement: Fundamental bimodal structure Synthetic Scaled synth(normalised) Synthetic 70 HARPS Scaled HARPSsynth • Agreement: Change in regime at ~20 M⊕: 60 frequency HARPS smoking gun of core accre6on: runaway gas 0.6 60 50 accre6on Mcore~Menve~10 M⊕ (but see also number Number 50 Bennet et al. 2021). 40 Number 0.4 40 Normalised 30 • Disagreement: Giant planets factor 2-3 too massive Raw 30 20 • Disagreement: Too few intermediate mass planets by 0.2 20 factor ~2-3 (planetary desert, Ida & Lin 2004). 10 100 100 101 102 103 104 too fast gas accre6on (cf. Nayakshin et al. 2019) 00 0 Msini [M ] 10 101 102 103 104 Msini [M ] Similar for gas accre6on rate derived from several 3D hydrodynamic models (Machida et al. 2010, D'Angelo et al. 2010, Bodenheimer et al. 2013) Possible explana6ons: low viscosity disks (Ginzburg & Chiang 2019a), magne6c regula6on (Batygin 2018, Cridland 2018), angular momentum barrier (Takata & Stevenson 1996), 3D circula6on (Szulagyi et al. 2014), …. Popula6on synthesis makes it possible to quan%fy discrepancies between theory and observa6ons.
Statistical comparison with Kepler survey c) d) NASA Kepler space telescope (transits) Synthetic t=5 Gyr e) Observed Mulders et al. (2019) Figure 5. Planet occurrence rates of the planet population synthesis model (abcd) compared to those of Kepler (e). Bins span equal areas in logaritmic area units (d log R d log P ) and roughly correspond to hot Jupiters, warm giants, super-earths, and • Agreement: rela6ve occurrence, except for (sub)- mini-Neptunes. The models systematically underestimate the occurrence of mini-Neptunes compared to the observed rates. Numbers: Bias corrected occurrence rates (Nb of planets in area / Nb of stars x 100) Neptunes (2-4 R⊕) • Disagreement: absolute occurrence: too high again (x 5) For a direct comparison, the end-to-end • Radius valley not clearly visible for high number of ini6al model should include the long term evolution embryos: Impact stripping? Water-rich planets? Enriched / internal structure calculation including atmospheric escape -> R and L/mag. envelopes?
Statistical comparison with direct imaging SPHERE@VLT SHINE GTO Actual detections & Synthetic population & survey (Vigan et al. 2020) sensitivity maps sensitivity maps 150 stars Plot by Clemence Fontanive cf. Nielsen et al. 2019 GPI SPHERE Very Large Telescope VLT +4.7 Observed: 5.8−2.8 % Frac6on of FGK stars w. planets +0.5 (M=1-75 MJ, a=5-300 AU) Synthe6c: 3.4−0.5 % • Agreements: overall frequency, mass-luminosity rela6on (β Pic b) • Distant giants in synthesis: Single, massive, eccentric planets from Probes very different kind of scanering events (see Marleau+2019b), mean eccentricity: 0.39 planets and a different • Disagreement: No HR 8799-like systems: 4 distant massive giants on observable (luminosity). rather circular orbits • Structured disks? Forma6on by gravita6onal instability?
4. Perspectives and conclusions
Comparisons with observations Andrews+2018 Initial Conditions: Probability Models of individual Global end-to-end form- distributions of disk properties processes ation & evolution model Disk gas mass From Disk dust mass Accretion, migration, … Disk properties planet properties observations Disk lifetime Draw and compute synthetic population New instrumentation better observational constraints Apply observational detection bias Predictions (going back to the full synthetic population) Observed population Stat. Comparison: Observable sub-population Model - Frequencies solution No match: change - Orbits, masses, radii, luminosities - Architecture, multiplicity Match found parameters, improve - Correlations model, reject model ….. One learns a lot even if a synthetic population does not match the observed one!
6 ? Perspectives and conclusions 4 20 2 0 0 2 4 6 8 10 20 40 60 80 100 2 4 6 8 10 20 Companion Mass [MJup ] Companion Mass 100 pc GAIA satellite Nancy Grace Roman satellite 50 pc Detectable 25 pc GAIA USE AM PLOT TO SHOW GAIA Data release ~2025 Data release ~2030 AND ROMAN Astrometric technique Microlensing Explore uncharted territory Expected yield: 20’000 technique say nb of giant planets! to 70’000 giant Expected yield: Detectable goals planets (!) 2000-3000 cold low- Roman mass planets Use: old map -> new map with america PLATO: transits ARIEL: atmospheric spectroscopy Observa6onally driven field: consider coming missions The future is bright regarding new sta6s6cal observa6onal constraints. Data release ~2030 Launch ~2029 Exquisite knowledge of planet mass distribu6on and demographics in giant and low-mass regime. Ideal to Transit technique Atmo. spectroscopy inves6gate mechanisms of gas and solid accre6on. Hab. zone planets Sta6s6cs of atmo. Blue lines: 5 σ detec+on limits for GAIA (Courtesy D. Segransan, Geneva Obs.) Temporal evolu6on composi6ons
Terra incognita as litmus test 2021 ~2030 Cross-checking the same theoretical model with population synthesis in many different and especially unexplored parameter spaces: Key to understand whether theory really captures the governing underlying physics and is not merely a sophisticate fit tweaked to explain already known observations. Much to do on the theoretical side: initial conditions & early phases, disk models (beyond α-models), hybrid pebble-planetesimal models, link formation-atmospheric composition, gas accretion,… Observing planet formation as it happens as a new direct constraint on planet formation
Conclusions •Population synthesis is a tool to compare theory and observation to improve understanding of planet formation • use full wealth of observational constraints • put detailed models to the test • see global statistical consequences: which processes are key? •Observational constraints on many processes • solid and gas accretion rate • N-body dynamics • orbital migration rate •See link between disk and planetary properties •Predict yield of future instruments/space missions •Continuously improving models • population syntheses depend on progress of formation theory as a whole • many new theoretical developments to test, many new obs. constraints to come
Resources Population synthesis review papers -Benz et al., Protostars & Planets VI, 691, 2014 -Mordasini et al., IJA, 201, 2015 -Mordasini, Handbook of Exoplanets, 143, 2018 DACE data base: Bern population synthesis models https://dace.unige.ch/evolution/index All NGPPS data publicly available via dedicated interac6ve online tool on DACE website Freely available toy population synthesis model http://nexsci.caltech.edu/workshop/2015/#handson
You can also read