Skip to content

Volatiles held under a closed crust

This tutorial shows how a crust that closes over the magma ocean holds the volatiles of the melt, and how they then leak out down to a floor. You run the same planet twice, once with the lid seal and once without it. Then you read when the seal closed, check when each volatile starts to leak, and compare the 2 atmospheres.

When the seal closes, the volatiles dissolved in the melt move to the {x}_kg_sealed columns, and the chemistry stops. While the seal holds, the held volatiles reach the atmosphere only by a leak with the relaxation time tau_leak, and by volcanism if crust_rate is above 0. The leak stops at the floor {x}_kg_eq: the mass that the melt dissolves at the pressure of the atmosphere. Volatiles held under a closed crust explains the rule.

Prerequisites

  • PROTEUS installed with bash install.sh (see the installation guide), with the Aragog interior, and its environment active (conda activate proteus).
  • FWL_DATA set; the first run downloads the interior data.
  • tomlkit and matplotlib; both come with pip install -e ".[develop]".

Run the commands from the PROTEUS folder. Run the Python blocks in one Python session, in the order of the steps. Each run took about 20 minutes on a workstation. The printed values come from runs with CALLIOPE 26.11.8.

Step 1: the configuration

The configuration is input/tutorials/tutorial_lid_seal.toml. An Earth-mass planet at 1 AU cools from a magma ocean under a grey (dummy) atmosphere, with the Aragog interior and the CALLIOPE chemistry. Its seal section is:

[outgas.lid_seal]
    enabled = true
    tau_leak = 1.0e5             # [yr]; delta_close stays at its default, 10 km

The seal closes when the crust is at least delta_close thick (default 10 km) and the top mesh cell holds less than phi_close of melt (default 0.3). Here no sub-grid lid runs, so the crust is the top run of solid mesh cells. The run stops at 140 kyr.

Step 2: run the planet with the seal

proteus start --offline -c input/tutorials/tutorial_lid_seal.toml

The output goes to output/tutorial_lid_seal/.

Step 3: run the same planet without the seal

The block writes a copy of the configuration with the seal off and its own output folder:

import tomlkit

cfg = tomlkit.parse(open('input/tutorials/tutorial_lid_seal.toml').read())
cfg['params']['out']['path'] = 'tutorial_lid_seal_off'
cfg['outgas']['lid_seal']['enabled'] = False
open('input/tutorials/tutorial_lid_seal_off.toml', 'w').write(tomlkit.dumps(cfg))
proteus start --offline -c input/tutorials/tutorial_lid_seal_off.toml

Step 4: when did the seal close?

import numpy as np
import pandas as pd

on = pd.read_csv('output/tutorial_lid_seal/runtime_helpfile.csv', sep=r'\s+')
off = pd.read_csv('output/tutorial_lid_seal_off/runtime_helpfile.csv', sep=r'\s+')
k = int(np.flatnonzero(on['seal_active'].values > 0.5)[0])
close = on.iloc[k]
print(f"closed at {close['Time']:.0f} yr, crust {close['seal_delta'] / 1e3:.1f} km, "
      f"top-cell melt fraction {close['seal_phi_top']:.1e}, mantle melt fraction {close['Phi_global']:.3f}")
print(f"held: H2O {close['H2O_kg_sealed']:.3e} kg, CO2 {close['CO2_kg_sealed']:.3e} kg")

The run printed:

closed at 94053 yr, crust 65.8 km, top-cell melt fraction 5.1e-04, mantle melt fraction 0.199
held: H2O 6.924e+18 kg, CO2 2.950e+17 kg

The seal closes when the top mesh cell freezes. The top cell of this mesh is 65.8 km thick, so the crust jumps from 0 to 65.8 km in one step, and any delta_close up to 65.8 km closes the seal at the same step. About a fifth of the mantle is still molten, and its dissolved water and CO\(_2\) are now held under the crust.

Step 5: when does each volatile start to leak?

At closure the held mass of each volatile is the mass dissolved in the melt. Without trapping, its floor counts the melt alone, so each held volatile starts on its floor. When the global melt fraction \(\Phi\) falls, the melt shrinks, the floor falls below the held mass, and the volatile leaks:

held = on.iloc[k:]
print(f"floor / held water on the first held step: {held['H2O_kg_eq'].iloc[1] / held['H2O_kg_sealed'].iloc[1]:.6f}")
for s in ('CO2', 'H2O'):
    leaks = held[f'{s}_kg_sealed'].diff() < 0
    first = held[leaks].iloc[0]
    print(f"{s} first leaks at {first['Time']:.0f} yr, Phi = {first['Phi_global']:.5f}")
leaks = held['H2O_kg_sealed'].diff() < 0
print(f"held / floor of water on the rows where it leaks: at least {(held['H2O_kg_sealed'] / held['H2O_kg_eq'])[leaks].min():.6f}")

The run printed:

floor / held water on the first held step: 0.999739
CO2 first leaks at 94135 yr, Phi = 0.19854
H2O first leaks at 94135 yr, Phi = 0.19854
held / floor of water on the rows where it leaks: at least 1.000261

On the first held step \(\Phi\) falls from 0.19873 to 0.19854, and water and CO\(_2\) leak from that step on. From then on the held water follows its floor down and stays above it.

Step 6: compare the 2 atmospheres

end_on, end_off = on.iloc[-1], off.iloc[-1]
print(f"H2O in the atmosphere at 140 kyr: seal {end_on['H2O_kg_atm']:.3e} kg, no seal {end_off['H2O_kg_atm']:.3e} kg")
print(f"surface pressure: seal {end_on['P_surf']:.3f} bar, no seal {end_off['P_surf']:.3f} bar")
print(f"held H2O: {end_on['H2O_kg_sealed']:.3e} kg, floor {end_on['H2O_kg_eq']:.3e} kg")
print(f"H2O in the melt without the seal: {end_off['H2O_kg_liquid']:.3e} kg")
print(f"mantle melt fraction without the seal: {end_off['Phi_global']:.3f}")

The run printed:

H2O in the atmosphere at 140 kyr: seal 2.056e+15 kg, no seal 2.021e+15 kg
surface pressure: seal 1.036 bar, no seal 1.063 bar
held H2O: 6.923e+18 kg, floor 6.920e+18 kg
H2O in the melt without the seal: 6.923e+18 kg
mantle melt fraction without the seal: 0.140

Before the closing step the 2 runs agree to the last digit. After it they part, but not by much. The melt is still 14.0 % of the mantle at 140 kyr, and in both runs almost all the water stays in the mantle. With the seal, the atmosphere holds 1.73 % more water than without the seal. Replays of the held span of this run, with one term switched off at a time, split the 1.73 % into 3 terms, in percentage points of the water in the atmosphere of the open run:

  • \(+1.86\): the held gas keeps the species it had at closure, while the open run turns CO into CO\(_2\) as the magma cools; the sealed atmosphere then holds less oxygen, the kept species lower its surface pressure by 1.77 %, and the water needs more mass in the atmosphere to reach the same partial pressure;
  • \(-0.07\): the finite leak time keeps this water held;
  • \(-0.07\): the rest.

The output above shows a surface pressure 2.6 % lower in total; the kept species alone lower it by 1.77 %. The closing temperature that the seal keeps for the CO\(_2\) law changes nothing here: near 1 bar that law does not depend on temperature.

The first term is a known limit of the frozen chemistry: at 140 kyr the sealed atmosphere holds 24.7 % more CO and 6.3 % less CO\(_2\) than the open run (Limits of the leak).

Over the run printed here and 5 starts whose initial superheat differs by up to \(2.5 \times 10^{-9}\) K, the closing time varies by 2.5 yr and the water in the atmosphere at 140 kyr by \(7.2 \times 10^{-4}\) % with the seal and by \(9.6 \times 10^{-4}\) % without it. The seal run and the open run of one start agree to the last digit before closure, so the seal effect is their difference at the same start: 1.730 % in each of the 6 starts. These starts run on the default mesh, with a 3000 m CMB cell; the spread on other meshes was not measured. tests/data/lid_seal/tutorial_ensemble.csv holds these end values.

The block plots the water in the atmosphere, and the held water against its floor:

import matplotlib.pyplot as plt

fig, (ax, ax2) = plt.subplots(2, 1, figsize=(7.5, 6), sharex=True)
ax.plot(on['Time'] / 1e3, on['H2O_kg_atm'], label='seal')
ax.plot(off['Time'] / 1e3, off['H2O_kg_atm'], '--', label='no seal')
ax.set_ylabel('H$_2$O in the atmosphere [kg]')
ax.legend(loc='upper left', bbox_to_anchor=(1.02, 1.0))
on_floor = held[held['H2O_kg_eq'] > 0]
ax2.plot(on_floor['Time'] / 1e3, on_floor['H2O_kg_sealed'] / on_floor['H2O_kg_eq'] - 1)
ax2.axhline(0.0, ls=':')
ax2.set_ylabel('held / floor - 1')
ax2.set_xlabel('Time [kyr]')
fig.tight_layout()
fig.savefig('output/tutorial_lid_seal/water.png', dpi=150)

Step 7: try it yourself

  • Set tau_leak to 1e2 and run again: the leak follows the floor more closely, and the water in the atmosphere at 140 kyr changes by less than 0.1 %.
  • Set crust_rate above 0 to add the volcanic leak; the columns {e}_kg_leak_volc and {e}_kg_leak_tau show the 2 parts in each step, and seal_volc_capped shows the steps where the crust production was cut to the melt mass. Both parts stop at the same floor.
  • Set delta_close to 0.3 m: on this mesh the seal closes at the same step, because a crust on the mesh is 0 or a whole cell. A thin crust closes the seal earlier only through the Aragog sub-grid lid.