(732d) Integrated Computational Screen and Design of Polymers for Post-Combustion Carbon Capture

Developing polymer membranes for cost-effective post-combustion carbon capture is challenging. Recent process modeling shows that for cost-effective post-combustion carbon capture using membrane separation technology, a polymer should have CO₂ permeance above 2000 GPU (10^−6 cm³(STP)/(cm² s cm Hg)) and CO₂/N₂ permeance selectivity above 25 ^[1] . Note that 1 GPU of permeance corresponds to 1 barrer (1×10^−10 cm³ (STP)· cm/cm² s cm Hg) of permeability if the selective polymer membrane thickness is assumed to be 1 mm (10^−6 m). A survey of the MSA polymer gas separation database ^[2] shows that only four sets of experimental data out of a total of ~ 1500 data sets exhibit CO₂ permeability above 2000 barrer and CO₂/N₂ selectivity above 25, and those polymers are glass at room temperature and as a result the physical aging of the glass significantly affects their performance.

To address the above challenges of screening and designing rubber polymers for carbon capture, we adopted an integrated approach, which includes database surveys, chem-informatics and machine learning, and molecular modeling. Specifically, we have web-scraped the on-line CROW ^[3] polymer database. This database includes important polymer physical properties, such as glass transition temperature (Tg), polymer density, and van-der Waals volume for the repeat unit, from which the polymer free volume fraction was estimated. There are 238 polymers whose physical properties were collected; 111 polymers are rubbers at room temperature. Furthermore, CO₂ and N₂ permeability values in many polymers obtained from the CROW database were estimated using the polymer Genome platform and machine-learning models ^[4] developed by Ramprasad and coworkers. Preliminary calculations show that out of the 111 rubbers, the polydimethylsiloxane (PDMS) polymers exhibit a CO₂ permeability of ~4000 barrer, which is reasonably close to the experimental data and our molecular simulation results. All other rubbers calculated so far exhibit CO₂ permeability much less than 1000 barer. Note that both our molecular simulations and the literature experimental data show that the PDMS rubber polymer exhibits a CO₂/N₂ permeability selectivity of 7-9, much less than 25.

To further functionalize PDMS rubber to increase CO₂/N₂ permeability selectivity while maintaining its high CO₂ permeability, we utilized the computational database, which we recently developed to characterize CO₂ interactions with about 202 different functional groups. We also utilized the simplified molecular-input line-entry system (SMILES) and the RDKit software package implanted in Python to generate molecular structures for the above 111 rubbers to quickly look through the branch-structures of the polymers to functionalize the PDMS polymer. Finally, molecular simulation results of CO₂ and N₂ solubility^[5,6], diffusivity, permeability, and permeability selectivity will be presented for the above functionalized PDMS polymers.

References:

1. Zoelle, R. Newby, “Performance and Cost Sensitivities for Post-Combustion Membrane Systems”, NETL CO₂ capture Technology Project Review Meeting, Pittsburgh, PA, August 13-17, 2018.
2. Membrane Society of Australasia (MSA) Polymer Gas Separation Membrane Database, “https://membrane-australasia.org/msa-activities/polymer-gas-separation-membrane-database/”, accessed and downloaded locally on Aug 23, 2018.
3. CROW Polymer Science, http://www.polymerdatabase.com/, web-scraped on August 15 2019.
4. Polymer Genome, “https://www.polymergenome.org/”
5. Wei Shi, Robert L. Thompson, Megan K. Macala, Kevin Resnik, Janice A. Steckel, Nicholas S. Siefert, and David P. Hopkinson, “Molecular Simulations of CO₂ and H₂ Solubility, CO₂ Diffusivity, and Solvent Viscosity at 298 K for 27 Commercially Available Physical Solvents”, Invited paper to the Chem. Eng. Data, DOI: 10.1021/acs.jced.8b01228, 2019
6. Wei Shi, E. J. Maginn, “Continuous Fractional Component Monte Carlo: An Adaptive Biasing Method for Open System Atomistic Simulations”, Chem. Theory Comput. 2007, 3, 1451-1463.