Input files and scripts to reproduce the results of Lesher et al. (2020): Iron isotope fractionation at the core-mantle boundary by thermodiffusion:
In order to run the models, you need to have ASPECT installed on your computer. Installation instructions can be found here. However, instead of the development version, you will need to install the branch of ASPECT that was used to run the models in the manuscript, which you can download from github with the following command
git clone https://github.com/jdannberg/aspect.git
(instead of the git clone command listed in the ASPECT manual installation instructions under point 3.3.3. Apart from that, you can follow the instructions exactly as listed in the manual.
All model outputs can be visualized using for example the data analysis and visualization application ParaView.
There are three sets of models we ran for this study:
The input files for these models are located in the the reproducing-the-experimental-setup folder. They can be used to reproduce the ASPECT models shown in Figure S5. There are three different input files, with their names giving the value of alpha as listed in the caption of Figure S5.
The models can be run with the command
/path/to/aspect/executable/in/build/directory input-file.prm
(where /path/to/aspect/executable/in/build/directory should be the path to the aspect executable on your system,
and input-file.prm should be replaced by the name of one of the three input files, for example SOR-27-alpha-2.8.prm.)
The input files for these models are located in the the 1D-analysis-of-fractionation-near-CMB folder. They can be used to reproduce the ASPECT models shown in Figure S8. These models assume that iron can only diffuse up to 10 km into the lowermost mantle, so in order to reproduce these models, you will have to change and recompile the code. The required changes are given in the patch file patch1.patch and can be applied by copying this file into your aspect source directory and then executing
patch -p1 < patch1.patch
You will then need to recompile ASPECT.
You can run the models using the run_all_models script.
The script assumes that you have added an alias for ASPECT in your .bashrc file, so that you can run ASPECT by typing aspect input-file.prm.
If that's not the case, you will either need to add the following line to the end of your ~/.bashrc file
alias aspect=/path/to/aspect/executable/in/build/directory
(replacing /path/to/aspect/executable/in/build/directory by the path to the aspect executable on your system; you will need to open a new terminal for this change to take effect)
or you will need to change line 28 in the run_all_models script
by replacing aspect with the full path to the aspect executable on your system.
Then you can run the models with the command
bash run_all_models
This will create one output folder for each of the 1-D models shown in the manuscript.
The input files for these models are located in the the 2D-fractionation-models folder.
For computing the fractionation of iron isotopes, the models assume that liquids from the core can infiltrate into
the mantle as long as they are above their solidus (but they also track the motion of thin layers with 10/50/100 km thickness
starting out just above the core-mantle boundary to illustrate where core liquid ends up in case it can not infiltrate into
the mantle over longer distances).
Consequently, in case you have made the changes decribed above in the section about running the 1-D model, you have to revert them
(for example by running the command git checkout ./source/simulator/assemblers/advection.cc in your ASPECT directory)
and you will have to recompile ASPECT.
There are two input files: The file fractionation_k8.5_D2.5e-9.prm reproduces the model shown in Figure S9 top (with a thermal conductivity k = 8.5 W/m/K and diffusivity DFe = 2.5e-9 m2/s). The file fractionation_k4_D1e-8.prm reproduces the model shown in Figure S9 bottom (with a thermal conductivity k = 4 W/m/K and diffusivity DFe = 1e-8 m2/s).
These models have more than 4 million cells and we ran them on 192 cores, so you will likely need access to a powerful computer to rerun these models.