Volume 47, Issue 4 e2019GL085476
Research Letter
Free Access

Recreating Giants Impacts in the Laboratory: Shock Compression of urn:x-wiley:grl:media:grl60159:grl60159-math-0002 Bridgmanite to 14 Mbar

Marius Millot

Corresponding Author

Marius Millot

Lawrence Livermore National Laboratory, Livermore, CA, USA

Correspondence to: M. Millot,

[email protected]

Search for more papers by this author
Shuai Zhang

Shuai Zhang

Lawrence Livermore National Laboratory, Livermore, CA, USA

Laboratory for Laser Energetics, University of Rochester, Rochester, NY, USA

Search for more papers by this author
Dayne E. Fratanduono

Dayne E. Fratanduono

Lawrence Livermore National Laboratory, Livermore, CA, USA

Search for more papers by this author
Federica Coppari

Federica Coppari

Lawrence Livermore National Laboratory, Livermore, CA, USA

Search for more papers by this author
Sebastien Hamel

Sebastien Hamel

Lawrence Livermore National Laboratory, Livermore, CA, USA

Search for more papers by this author
Burkhard Militzer

Burkhard Militzer

Departments of Earth and Planetary Science and Astronomy, University of California Berkeley, Berkeley, CA, USA

Search for more papers by this author
Dariia Simonova

Dariia Simonova

Bayerisches Geoinstitut, University of Bayreuth, Bayreuth, Germany

Search for more papers by this author
Svyatoslav Shcheka

Svyatoslav Shcheka

Bayerisches Geoinstitut, University of Bayreuth, Bayreuth, Germany

Search for more papers by this author
Natalia Dubrovinskaia

Natalia Dubrovinskaia

Material Physics and Technology at Extreme Conditions, Laboratory of Crystallography, University of Bayreuth, Bayreuth, Germany

Search for more papers by this author
Leonid Dubrovinsky

Leonid Dubrovinsky

Bayerisches Geoinstitut, University of Bayreuth, Bayreuth, Germany

Search for more papers by this author
Jon H. Eggert

Jon H. Eggert

Lawrence Livermore National Laboratory, Livermore, CA, USA

Search for more papers by this author
First published: 21 January 2020
Citations: 18

Abstract

Understanding giant impacts requires accurate description of how extreme pressures and temperatures affect the physical properties of the constituent materials. Here, we report shock experiments on two polymorphs of MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0003: enstatite and bridgmanite (perovskite) crystals. We obtain pressure-density shock equation of state to 14 Mbar and more than 9 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0004, a 40% increase in density from previous data on MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0005. Density-functional-theory molecular dynamics (DFT-MD) simulations provide predictions for the shock Hugoniot curves for bridgmanite and enstatite and suggest that the Grüneisen parameter decreases with increasing density. The good agreement between the simulations and the experimental data, including for the shock temperature along the enstatite Hugoniot reveals that DFT-MD simulations reproduce well the behavior of dense fluid urn:x-wiley:grl:media:grl60159:grl60159-math-0006. We also reveal a high optical reflectance indicative of a metal-like electrical conductivity which supports the hypothesis that magma oceans may contribute to planetary magnetic field generation.

Key Points

  • Bridgmanite was shock compressed to the conditions of giant impacts that dominated the final phase of the solar system formation
  • DFT molecular dynamics simulations reproduce well the experimental data under such extreme P-T conditions relevant for giant impact models
  • Dense fluid urn:x-wiley:grl:media:grl60159:grl60159-math-0008 exhibits a metal-like electrical conductivity which indicates that magma oceans may generate planetary magnetic field

Plain Language Summary

Deciphering the evolution of the early Earth requires a detailed understanding of the history of our planet formation and evolution. Much like for other planets in the solar system and beyond, giant impacts are thought to have played a key role in the Earth history including the formation of the moon and the intense climatic perturbations leading to the Cretaceous-Paleogene extinction event. Computer simulations of giant impact are now becoming increasingly accurate thanks to ever-growing supercomputing capabilities worldwide. Here we report new shock wave experiments on two different kinds of the Earth mantle's most abundant mineral MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0007, together with simulations based on quantum theory of condensed matter. We find that under intense shockwave compression of several million atmospheres, shock-induced heating and compression together transform the rocky minerals into dense, shiny fluid able to conduct electrical current and therefore perhaps contribute to magnetic field generation by dynamo effect in the early stages of the evolution of rocky planets and exoplanets.

1 Introduction

The high-energy collisions between planetary bodies that are called giant impacts are thought to play a key role in the final phase of the formation of planetary systems such as our solar system and could have contributed to forming the moon (Asphaug, 2014; Stevenson, 1987) and Jupiter's diluted core (Liu et al., 2019) and may contribute to explain some of Uranus oddity (Kegerreis et al., 2018).

Laser-driven shock compression now enables us to routinely recreate, in the laboratory, the megabar (1 Mbar = 100 GPa) pressure and thousand of degrees temperature conditions that occur during and in the aftermath of these events. We can therefore perform experiments to study how such extreme conditions modify the properties of typical planetary constituents such as iron (Kraus et al., 2015), periclase (McWilliams et al., 2012), and silicate minerals (Bolis et al., 2016; Fratanduono et al., 2018; Hicks et al., 2006; Millot et al., 2015; Root et al., 2018). High-precision experiments and complementary numerical simulations also provide essential benchmarks for developing equation of state (EOS) models of the constituent materials. Such EOS models are essential for the numerical simulations of event such as the moon forming impact on Earth which is thought to have included a Mars-sized terrestrial protoplanet colliding into the proto-Earth with a velocity near 10 km/s (Asphaug, 2014; Stevenson, 1987).

Several groups have characterized the thermodynamics of MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0009 at Mbar pressures using static compression (Hirose et al., 2017) and gas-guns drivers (Akins et al., 2004; Deng et al., 2008; Gong, 2004; Mosenfelder et al., 2009). In particular, shock temperature and density measurements as a function of shock pressure were interpreted as indicative of melting (Akins et al., 2004), but some ambiguity remains as no signatures as strong as those observed for SiO urn:x-wiley:grl:media:grl60159:grl60159-math-0010 (Millot et al., 2015) were found. Bolis et al. (2016) later reported shock temperature along the locus of accessible shock states, that is, the Hugoniot curve for MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0011 glass with laser-driven shock compression. Fratanduono et al. (2018) further documented the thermodynamical properties of the dense fluid by reporting pressure-density-temperature and sound speed measurements up to 8 Mbar along the enstatite Hugoniot. This most recent work also provided constraints on the melt transition along the enstatite Hugoniot and showed that the adiabats and melt boundary of fluid MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0012 are likely shallower than predicted by DFT-MD (Stixrude, 2014).

Here, we report additional experimental data to more extreme conditions representative of the giant impacts that dominated the final phase of the solar system formation, as well as DFT-MD EOS data that are found to be in good agreement with the experiments and allow us to show that the Grüneisen parameter decreases with increased density above 5 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0013, in contrast with previous models and simulations between 2.5 and 5 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0014 (Mosenfelder et al., 2009; Stixrude & Karki, 2005)

2 Experimental and Numerical Methods

2.1 Laser Shock Experimental Configuration

Bridgmanite samples were synthesized in a multianvil apparatus by a direct reaction between finely powdered MgO and SiO urn:x-wiley:grl:media:grl60159:grl60159-math-0019 at high-pressure high-temperature (HPHT) conditions (see details in the supporting information). As in previous experiments (Fratanduono et al., 2018; Lazicki et al., 2017; Millot et al., 2015; Root et al., 2018), the targets used in the present experiments contained a planar package with a urn:x-wiley:grl:media:grl60159:grl60159-math-002050  urn:x-wiley:grl:media:grl60159:grl60159-math-0021m polyimide (Kapton) ablator, a urn:x-wiley:grl:media:grl60159:grl60159-math-002250  urn:x-wiley:grl:media:grl60159:grl60159-math-0023m urn:x-wiley:grl:media:grl60159:grl60159-math-0024-quartz reference plate having a 3  urn:x-wiley:grl:media:grl60159:grl60159-math-0025m thick Au layer deposited on top of a 100 nm Ta coating, a 100–140  urn:x-wiley:grl:media:grl60159:grl60159-math-0030m bridgmanite crystal and a second urn:x-wiley:grl:media:grl60159:grl60159-math-0031 40  urn:x-wiley:grl:media:grl60159:grl60159-math-0032m quartz plate having an antireflection coating (see Figure 1).

Details are in the caption following the image
Experimental configuration. (a) Cylinder of polycrystalline MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0015 bridgmanite synthesized in run H4538. (b) Sample plate extracted from run H4538 and polished to 136  urn:x-wiley:grl:media:grl60159:grl60159-math-0016m thick ( urn:x-wiley:grl:media:grl60159:grl60159-math-0017300  urn:x-wiley:grl:media:grl60159:grl60159-math-0018m lateral size, used for shot s25219). Picture is taken in transmission with a white background. (c) Target sketch for the laser-driven shock experiments at the Omega EP Laser. (d) Measurement concept when the shock wave is strong enough to transform the unshocked material into a good optical reflector: The VISAR directly tracks the shock velocity.

One beam of the Omega EP Laser Facility at the Laboratory for Laser Energetics was used to drive the sample with urn:x-wiley:grl:media:grl60159:grl60159-math-0033700–1,000  urn:x-wiley:grl:media:grl60159:grl60159-math-0034 of 351 nm UV light deposited in a flat-top 1 ns pulse. Beam smoothing and a distributed phase plate created an eight-order super Gaussian laser intensity distribution with a 1,100 urn:x-wiley:grl:media:grl60159:grl60159-math-0035 diameter in the focal plane. The primary diagnostic is a line-imaging velocity interferometer system for any reflector (VISAR) operating at 532 nm and using a Rochester Optical Streak System (ROSS) streak camera in each of its two channels (Celliers et al., 2004). We used fused silica etalon stacks totaling 18.1985 and 7.3657 mm giving a vacuum velocity per fringe sensitivity of 2.7361 and 6.7601 km/s/fringe. Sweep windows were 27 and 15 ns; 2 GHz timing combs were recorded on each streak record and used to accurately determine the time versus pixel relationship.

A one-dimensional fast Fourier transform algorithm was used to determine the space-time maps of the VISAR phase-shift from the streaked images. Due to the presence of spurious ghost fringes generated at the bridgmanite/glue/quartz interfaces we used a simple heuristic ghost fringe removal algorithm (Millot et al., 2018) to subtract stationary fringes from the raw images (see Figure 2).

Details are in the caption following the image
VISAR data Raw (Top) and corrected (Middle) records for VISAR channel A for shot 25219 showing the unambiguous signatures of an unsupported, reflecting shock front. The shock arrival (breakout) in the bridgmanite is highlighted by the red vertical arrow. The ghost-fringe removal algorithm (Millot et al., 2018) reveals more clearly the fringes associated with the shock front traveling in the bridgmanite sample. (Bottom) Inferred shock velocity history with channel A(Red) and B (Blue). Shaded colors represent systematic uncertainty estimated as 5% of the velocity per fringe. A linear fit (thick green lines) of the velocity right before and after the shock breakout event urn:x-wiley:grl:media:grl60159:grl60159-math-0026 is used to determine urn:x-wiley:grl:media:grl60159:grl60159-math-0027 and urn:x-wiley:grl:media:grl60159:grl60159-math-0028. The dotted vertical line indicates urn:x-wiley:grl:media:grl60159:grl60159-math-0029 with thin black lines representing the 50 ps timing uncertainty.The jump in shock velocity near 2.5 ns is due to a wave reverberation in the Au layer.

Similar target design, target fabrication, laser configuration, and diagnostics were used to measure the properties of MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0036 along the enstatite Hugoniot using natural crystals having the following chemical formula: (Mg urn:x-wiley:grl:media:grl60159:grl60159-math-0037Fe urn:x-wiley:grl:media:grl60159:grl60159-math-0038Al urn:x-wiley:grl:media:grl60159:grl60159-math-0039) (Si urn:x-wiley:grl:media:grl60159:grl60159-math-0040Al urn:x-wiley:grl:media:grl60159:grl60159-math-0041)O urn:x-wiley:grl:media:grl60159:grl60159-math-0042. For details, see supporting information and Fratanduono et al. (2018).

2.2 Velocimetry Analysis

The raw VISAR data (Figure 2) are similar to previous experiments on diamond (Hicks et al., 2008), silica (Hicks et al., 2006; Millot et al., 2015), and enstatite (Fratanduono et al., 2018). In the three experiments reported here, the VISAR data indicate that the unsupported shock traveling in the target package is initially strong enough to transform both the SiO urn:x-wiley:grl:media:grl60159:grl60159-math-0043 reference material and the MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0044 sample into good optical reflectors (Millot, 2016). This is evident by the strong time dependence of the VISAR fringe shift with time, characteristic of an unsupported shock (see Figure 2). The disappearance of the fringes emanating from the shock front (after ghost fringe removal) is also evident near the end of the shock transit in the bridgmanite sample. This happens when the shock decays below the pressure required to transform the bridgmanite into a reflective dense fluid.

We show on Figure 2 the velocity history urn:x-wiley:grl:media:grl60159:grl60159-math-0045 of the shock during its transit through the quartz reference plate and the bridgmanite sample (where we can detect fringes from the reflection at the shock front). We determine the values urn:x-wiley:grl:media:grl60159:grl60159-math-0046 and urn:x-wiley:grl:media:grl60159:grl60159-math-0047, respectively just prior to and after the instant urn:x-wiley:grl:media:grl60159:grl60159-math-0048 when the shock breaks through the quartz/bridgmanite interface by linearly fitting urn:x-wiley:grl:media:grl60159:grl60159-math-0049 over 350 ps before and after urn:x-wiley:grl:media:grl60159:grl60159-math-0050 and extrapolating to urn:x-wiley:grl:media:grl60159:grl60159-math-0051. To obtain the shock velocity (Celliers et al., 2004), we used urn:x-wiley:grl:media:grl60159:grl60159-math-0052 (quartz) = 1.547 and urn:x-wiley:grl:media:grl60159:grl60159-math-0053 (bridgmanite) =  urn:x-wiley:grl:media:grl60159:grl60159-math-0054, for which we calculated the average of the three indices reported at 514 nm by Yeganeh-Haeri (1994) and assumed negligible dispersion.

Using the previously calibrated reflectance of the shock front in the quartz reference (Hicks et al., 2006; Brygoo et al., 2015), we also infer the shock front reflectivity in the bridgmanite using the amplitude of the VISAR fringes as described in Millot et al. (2018).

2.3 EOS Simulations with Quantum Molecular Dynamics

We performed EOS molecular dynamics simulations based on density functional theory (DFT) with the local density approximation (LDA) using the Vienna Ab-initio Simulation Package (VASP) (Kresse & Furthmuller, 1996). We compute the internal energy and the pressure of dense MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0055 over a grid of densities and temperatures spanning an extended range compared to previous calculations (Militzer, 2013) : 5.4–9.8 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0056 and 5,000–35,000 K. We also calculate urn:x-wiley:grl:media:grl60159:grl60159-math-0057 and urn:x-wiley:grl:media:grl60159:grl60159-math-0058 at ambient conditions for the three starting materials considered here: glass with urn:x-wiley:grl:media:grl60159:grl60159-math-0059g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0060, enstatite single crystals with urn:x-wiley:grl:media:grl60159:grl60159-math-0061g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0062 and bridgmanite crystals with urn:x-wiley:grl:media:grl60159:grl60159-math-0063 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0064. We approximate urn:x-wiley:grl:media:grl60159:grl60159-math-0065 for the glass by the energy of enstatite at the same density. We then compute the Hugoniot curves shown on Figures 3 and 4 by solving the Rankine-Hugoniot equation: urn:x-wiley:grl:media:grl60159:grl60159-math-0066 and considering that the shock states are either within the solid phase or within the fluid phase (labeled solid and liquid in the figure legends). Details about the simulations, tabular EOS values and a comparison with the simulated Hugoniot curves published in Militzer (2013) can be found in the supporting information. EOS results from first-principles simulation at higher P and T are available in González-Cataldo (2020).

Details are in the caption following the image
(Top) Shock velocity urn:x-wiley:grl:media:grl60159:grl60159-math-0067 versus particle velocity urn:x-wiley:grl:media:grl60159:grl60159-math-0068. (Bottom) Pressure urn:x-wiley:grl:media:grl60159:grl60159-math-0069 versus density urn:x-wiley:grl:media:grl60159:grl60159-math-0070 for various polymorphs of MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0071. The new data collected with bridgmanite single crystals (orange circles) are compared with previous gas-gun data for polycrystalline bridgmanite-majorite assemblages from Mosenfelder et al. (2009) (orange squares), as well as data for polycrystalline bridgmanite ( urn:x-wiley:grl:media:grl60159:grl60159-math-0072 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0073) (orange diamonds) from Deng et al. (2008). Green, cyan, and purple squares represent data with lower density polymorphs including enstatite and glass reported in Mosenfelder et al. (2009), while the green triangles are laser-driven data from Fratanduono et al. (2018). Hugoniot curves calculated from the DFT-MD simulations with initial densities urn:x-wiley:grl:media:grl60159:grl60159-math-0074 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0075 (glass), urn:x-wiley:grl:media:grl60159:grl60159-math-0076 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0077 (enstatite), and urn:x-wiley:grl:media:grl60159:grl60159-math-0078 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0079 (bridgmanite) and for shocks states either in the solid (thin solid lines) or fluid (thick solid lines) phases are also shown with colors representing the initial density.
Details are in the caption following the image
Shock Temperature urn:x-wiley:grl:media:grl60159:grl60159-math-0090 versus shock pressure urn:x-wiley:grl:media:grl60159:grl60159-math-0091 for various polymorphs of MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0092. We report here the full range of data collected in the study described in Fratanduono et al. (2018) along the enstatite Hugoniot. Hugoniot curves calculated from the DFT-MD simulations with initial densities urn:x-wiley:grl:media:grl60159:grl60159-math-0093 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0094(glass), urn:x-wiley:grl:media:grl60159:grl60159-math-0095 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0096 (enstatite), and urn:x-wiley:grl:media:grl60159:grl60159-math-0097 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0098 (bridgmanite) and for shocks states either in the solid (thin solid lines) or fluid (thick solid lines) phases are also shown with colors representing the initial density together with predicted melting lines from Belonoshko et al. (2005); Stixrude and Karki (2005); de Koker and Stixrude (2009) and determined from various experimental data in Fratanduono et al. (2018).

3 Results and Discussion

The observed discontinuity in the shock velocity originates from the difference in shock impedance between the reference material and the sample. Relying on previous measurements of the shock compressibility of urn:x-wiley:grl:media:grl60159:grl60159-math-0080 along the Hugoniot starting from urn:x-wiley:grl:media:grl60159:grl60159-math-0081quartz (Knudson & Desjarlais, 2013) we can apply the impedance matching procedure (Millot et al., 2018) to determine the compressibility of the sample using the measured jump in shock velocity at urn:x-wiley:grl:media:grl60159:grl60159-math-0082, and knowledge of the sample and reference initial densities. We used the Hugoniot fit from Brygoo et al. (2015) and the Mie-Grüneisen reshock model from Hicks et al. (2008) using urn:x-wiley:grl:media:grl60159:grl60159-math-0083 and a constant Grüneisen parameter urn:x-wiley:grl:media:grl60159:grl60159-math-0084 for the quartz reference and urn:x-wiley:grl:media:grl60159:grl60159-math-0085g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0086 for the bridgmanite sample. Experimental uncertainties were propagated with a Monte-Carlo routine with 5,000 iterations.

We show the results of the impedance matching procedure on Figure 3 (top), together with previous results on other polymorphs of urn:x-wiley:grl:media:grl60159:grl60159-math-0087. The new data reveal shock velocities higher than previous experiments on enstatite (Fratanduono et al., 2018). The corresponding particle velocities near 13 km/s are about three times higher than previous experiments with bridgmanite-rich starting materials (Deng et al., 2008; Mosenfelder et al., 2009) and confirm that we have reached the regime of the giant impacts that dominated the final phase of the solar system formation.

As expected, the new data lie at higher shock speed for a given particle velocity than the previous data collected along the Hugoniot of enstatite (Fratanduono et al., 2018). The new data also lie close to the Hugoniot curve for bridgmanite calculated with our DFT-MD simulations. Because the predicted Hugoniot curves in the solid and fluid phase lie very close to each other for a given starting material, the volume changes upon melting along the Hugoniot are expected to be rather small and any discontinuities in urn:x-wiley:grl:media:grl60159:grl60159-math-0088- urn:x-wiley:grl:media:grl60159:grl60159-math-0089 extremely challenging to identify experimentally.

We show all these results in the pressure-density space on Figure 3 (bottom). Our data near urn:x-wiley:grl:media:grl60159:grl60159-math-0099g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0100 represent a measurement of the compressibility of dense fluid MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0101 up to roughly threefold compression compared with the ambient condition solid polymorphs and extend previously available data by about 40% in density. As a comparison, 14 Mbar is the expected pressure at the core-mantle boundary of a 10 urn:x-wiley:grl:media:grl60159:grl60159-math-0102 super-Earth (Wagner et al., 2012). The Hugoniot curves obtained from the DFT-MD EOS simulations are found to be in good agreement with the previous lower pressure data obtained from gas-gun shock experiments on enstatite (Fratanduono et al., 2018), bridgmanite, and bridgmanite-majorite assemblages (Deng et al., 2008; Mosenfelder et al., 2009).

We also show in Figure 4 the full range of the shock temperature data along the enstatite Hugoniot collected in the previous work by Fratanduono et al. (2018). An excellent agreement is found between the experimental data and the simulated Hugoniot curve, up to 20 000 K. This provides additional confidence into the prediction for the bridgmanite Hugoniot which suggests that melting should be observed between 4 and 5.3 Mbar (400 to 530 GPa) relying on an extrapolation of the experimentally derived melting curve from Fratanduono et al. (2018), or between 4.7 and 5.8 Mbar if the predicted melting curve from Belonoshko et al. (2005) is more accurate.

The comparison between experiments and simulations up to 9 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0106 indicates that DFT-MD can accurately capture the thermodynamic properties of such a multicomponent system at extreme compression. The current work thus contributes to expand the range over which such experimental data can be used to benchmark the various theoretical and semiempirical EOS models that are used to simulate and better understand planetary formation and evolution including giant impacts. For example, many impact simulations use simplified EOS models such as the Mie-Grüneisen model. For example Hosono et al. (2019) explored the consequence of the observation that—in contrast with the commonly observed behavior in the solid phase—the Grüneisen parameter increases with increased density (Mosenfelder et al., 2009; Stixrude & Karki, 2005). Using a combination of experiments and DFT-MD simulations suggests that this behavior is in fact limited to between 2.5 and 5 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0107 and urn:x-wiley:grl:media:grl60159:grl60159-math-0108 at higher density up to 9 g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0109(see Figure S1).

We also show the shock front reflectivity along the bridgmanite Hugoniot in Figure 5, together with data collected, but not reported by Fratanduono et al. (2018) along the Hugoniot of enstatite. Note that, because of the imperfect transparency of the synthesized crystals, the values reported here for bridgmanite should be considered as lower bound (see Figure 1b, we used the first quartz plate as the reference). Comparing the data along the Hugoniot curves for the enstatite and bridgmanite crystals reveals that the onset of significant electronic conductivity is mostly temperature driven: the reflectivity exceeds 10–15% near urn:x-wiley:grl:media:grl60159:grl60159-math-0110 = 17–18 km/s for enstatite and urn:x-wiley:grl:media:grl60159:grl60159-math-0111 = 21–22 km/s for bridgmanite which correspond to very different pressures and densities but similar temperatures near 15,000–20,000 K

Details are in the caption following the image
Shock front reflectivity urn:x-wiley:grl:media:grl60159:grl60159-math-0103 as a function of shock velocity urn:x-wiley:grl:media:grl60159:grl60159-math-0104 for various polymorphs of MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0105. The new data collected with bridgmanite single crystals (orange circles, shots 25219 and 25216) at 532 nm are compared with previous laser-driven data along the enstatite Hugoniot collected in our previous study (Fratanduono et al., 2018).

The observed high reflectivity is very likely a signature of a strong increase in the imaginary part of the refractive index leading to optical conductivity values in excess of 1,000 S/cm near a photon energy of 2 eV (the VISAR probe is at 532 nm). This is similar to previous observations in dense fluids along the Hugoniot of MgO (McWilliams et al., 2012) and SiO2 (Hicks et al., 2006; Millot et al., 2015). This is also consistent with DFT-MD simulations (Qi et al., 2015; Soubiran & Militzer, 2018) indicating substantial electronic conductivity as the cause for the observed high reflectivity. Such a high electronic conductivity in dense fluid silicates could contribute to self-sustained magneto-hydro-dynamos within the primordial magma oceans of large terrestrial planets (Soubiran & Militzer, 2018). It could also reveal that dense fluid silicates have a high thermal conductivity. Such metallic-like properties could significantly alter the modeling of the aftermath of planetary collisions by affecting thermal exchange and inducing faster cooling rates for dense fluid silicates.

4 Summary and Discussion

We report novel experimental data on the shock compression of dense fluid MgSiO urn:x-wiley:grl:media:grl60159:grl60159-math-0112 using specially synthesized samples of bridgmanite. Ultrafast Doppler VISAR velocimetry yields shock EOS data up to 14 Mbar. Complementary computer simulations show that DFT-MD can accurately capture the thermodynamic properties of such a multicomponent system at extreme compression up to 9  g/cm urn:x-wiley:grl:media:grl60159:grl60159-math-0113.

Such experimental and numerical data on the shock compressibility and the associated temperature increase are essential to build improved EOS models such as the M-ANEOS model for urn:x-wiley:grl:media:grl60159:grl60159-math-0114 developed by Melosh (2007). Using such a model, one can then compute the entropy along the Hugoniot and—relying on the assumption that decompression is isentropic—calculate the pressure-density-temperature unloading path (Kraus et al., 2012). If the decompression isentrope intersects the liquid-vapor dome on the phase diagram, the expanding material will phase-separate into liquid and vapor. For example, Kraus et al. (2015) predict a transition from no vaporization to partial vaporization to occur as the speed of an iron-rich impactor increases from urn:x-wiley:grl:media:grl60159:grl60159-math-011515 to 20 km/s. We hope that the new data reported here will stimulate experimental and theoretical studies to provide similar estimates for urn:x-wiley:grl:media:grl60159:grl60159-math-0116-rich proto-planets. This is important because the vaporization fraction affects how far material is distributed after the impact as the generated vapor has a much larger volume and sets the power of the expansion.

Our experimental and numerical data could also directly be used to benchmark the various EOS models being used by the giant impact science community. Reducing the differences in the level of sophistication and accuracy for the EOS modeling could be very valuable and offer a better grasp on the underlying mechanisms at play in the fascinating phenomena unveiled in recent numerical simulation studies that aim to shed light on planetary collisions and their role in the birth and evolution of the solar system (Hosono et al., 2019; Kegerreis et al., 2019; Lock & Stewart, 2019, 2017; Nakajima & Stevenson, 2015; Pahlevan & Stevenson, 2007).

Acknowledgments

Authors gratefully acknowledge C. Davis, J. Emig, E. Folsom and R. Posadas Soriano for target preparation and the Omega Laser Facility management, staff and support crew for excellent shot and diagnostic support as well as insightful discussions with Sarah Stewart and Miki Nakajima. Prepared by LLNL under Contract DE-AC52-07NA27344. Partially supported by LLNL LDRD project 19-ERD-031 and LLNL HED Science Center. ND and LD received support from the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG, project No. DU 954/11-1) and the Federal Ministry of Education and Research, Germany (BMBF, grant No. 05K19WC1). BM received support through the DOE-NNSA grant DE-NA0003842. Omega shots were allocated through the LLE Laboratory Basic Science program.

    Data Availability Statement

    Data from this paper are available at https://doi.org/10.5061/dryad.z08kprr8r