Water plays an essential role in environmental chemistry, but when thousands of organic species are involved highly-detailed thermodynamic modeling can become problematic if not altogether impractical, due to limited species information. The BAT model was built to correct for this limitation as it is a water-sensitive organic species thermodynamic model that can adapt to different levels of chemical information. The BAT modeling framework describes a mixture of organics with water as binary mixtures of one organic and associated water.
For gas-particle partitioning, we use a non-ideal Volatility Basis Set (VBS) formulation that includes activity coefficients and accounts for liquid-liquid equilibrium. Our model is capable of using both detailed molecular structures as well as bulk properties like O:C, molar mass, and component volatility classes.
If you use the model please cite: Gorkowski, K., Preston, T. C., and Zuend, A.: Relative-humidity-dependent organic aerosol thermodynamics via an efficient reduced-complexity model, Atmos. Chem. Phys., 19, 13383–13407, https://doi.org/10.5194/acp-19-13383-2019, 2019.
The source code was written in MATLAB 2018b, though it will likely run with past and future versions. The source code is in Matlab_source_code, which contains update_Matlab_paths.m function that should be run to add the source folder to your MATLAB path.
In addition to the source code, a standalone executable file has been compiled and can run the model if you have no MATLAB license.
The Windows installation file is BAT_installer_v1_web.exe, which needs the freely available MATLAB Runtime R2018b engine. The installation file will download the respective runtime engine. You can also install the runtime engine independently from MathWorks.
Then run the installed program, and a file dialog box will appear. Then select one of the example input files, e.g., simple_input_single_run.txt.
The executable is expecting is a comma delimited text file, with one or more simulation input blocks.
Each input block starts with the run name, which is used when writing the output file.
run name=aPsoa_t1
The next line contains the refinement method for the BAT model, which is different from the VBS model. The BAT_refinement_mode can only be interpolate, different methods can be added. Note, this is only used when the water activity is above VBSBAT_options.BAT_refinement_aw=0.9 and the error in water activity is above VBSBAT_options.BAT_refinement_tolerance=1e-06
BAT_refinement_mode=interpolate
Next, are the VBS+BAT options, which have predefined run modes. You have to start with a predefined run mode, and then declare any changes in subsequent lines. The run mode options are default (used in Gorkowski et al. (2019)), robust which takes longer and provides a check on how well the neural networks are doing, NN only uses only the neural networks without refinement, and beta only which forces a single organic-rich phase.
VBSBAT_options.run_mode_used=default ,{default,robust,NN only,beta only}
Below this, you can add your modifications, like turning off the graphs.
VBSBAT_options.plot_PM=no ,{yes, no : option to make simple output graph}
The complete option list and settings used will be generated after each simulation run in a separate text file, like VBSBAT_input_used_aPsoat1.txt.
Then set the 'dry' water activity used to calculate the Csat_j, if only C_liquid and Cstar are given.
calculate_Csat_j_with_aw=0
Now water activity is used as a trigger word to indicate the next line is a list of water activities used to run this mixture and that the VBSBAT_options have ended. The water activity values can have non-uniform spacing, but should be monatomic.
water activity 0.9999,0.99,0.985,0.98,0.97,0.96,0.95,0.9,0.8,0.7,0.6,0.5,0.4,0.3,0.15,0.050.01,0.001,
The next trigger word is system properties, which is used to indicate the reading in of the organic composition. Note, currently the BAT functional group must have a space after and before the separating comma. The optional Name IS NOT OPTIONAL and does nothing currently, but it will be used in the future for more detailed output files. We are working on improving the input file reader.
system properties,
M (g/mol), O:C, H:C , eff. Csat_j (ug/m3), Ctotal_j (ug/m3) , BAT functional group, optional Name, optional Cstar (ug/m3), optional C^liquid (ug/m3),
2.00E+02,4.00E-01,1.60E+00,5.74E+03,8.79E+00, hydroperoxideSOA ,C107OOH,8620.171693,0.007057093,
1.88E+02,4.44E-01,1.78E+00,0.00E+00,3.98E+00, hydroperoxideSOA ,C97OOH,522.7659518,0.052085067,
2.16E+02,5.00E-01,1.60E+00,0.00E+00,1.13E+00, hydroperoxideSOA ,C108OOH,231.757194,0.032911505,
An empty line is used to separate simulation input blocks, and next simulation input block is triggered by a line containing run name
The figures presented in Gorkowski et al. (2019) can be reproduced using the source code in paper_figures. The data shown in the paper and the fit parameters are compiled in the Excel workbook file.