Project Info

Developing machine learning (ML) models to screen porous materials for CO2 capture from air

Diego Gomez-Gualdron
dgomezgualdron@mines.edu

Project Goals and Description:

In this project, the student will be looking to develop ML models to selectively predict CO2 adsorption under humid conditions to screen a database of 1 million porous materials for CO2 capture from air (where H2O is ever-present and CO2 is at dilute conditions). ML training will leverage adsorption data already obtained in the group for 50 thousands materials. The trained model will be then used for screening the larger database. The project is interesting because:
  1. With atmospheric CO2 levels being already above the 400 ppm threshold, ways to reduce atmospheric CO2 concentration are needed
  2. There is potential to discover better adsorbents for CO2 than other adsorbents currently piloted in industry, such as the CALF-20 metal-organic framework used by Svante.
  3. The work will leverage recent advancements in machine learning by my group (we have developed a machine readable representation of our materials that we hypothesize will make ML training more efficient) and others to train model and facilitate material discovery
  4. High probability for the undergrad to co-author a peer-reviewed research paper.

More Information:

Grand Challenge: Develop carbon sequestration methods.
-Our first effort to predict CO2 capture can be found here: https://pubs.acs.org/doi/10.1021/acs.chemmater.8b02257 -Our last effort on machine readable material representation and database screening can be found here: https://pubs.acs.org/doi/epdf/10.1021/acsami.4c11396 -A cool read about computational material discovery for CO2 capture can be read here:

Primary Contacts:

Diego A Gomez Gualdron: dgomezgualdron@mines.edu

Student Preparation

Qualifications

  • Scientific curiosity, thirst for knowledge, and passion for finding answers to scientific mysteries (required)
  • Finding the idea of working with computers exciting and entertaining (required)
  • A desire to make a positive impact on environment via scientific research (required)
  • A basic idea of how python works as expected from Mines introductory programming course (desirable but not required)
  • A cursory understanding of thermodynamics and "phase equilibrium (desirable but not required).
  • A desire or curiosity for pursuing graduate studies (desirable but not required)

TIME COMMITMENT (HRS/WK)

Can fluctuate week to week but 5 hours per week on average

SKILLS/TECHNIQUES GAINED

  • Understanding of thermodynamics and "phase equilibrium
  • Python knowledge to handle data and train machine learning models.
  • Foundational research skills and experience applying the scientific method.
  • Chemistry knowledge about chemical and physical interactions
  • Thermodynamic knowledge about phase equilibrium
  • Ability to work with supercomputers

MENTORING PLAN

-Mentoring assisted by grad student. -Mentoring by grad student (on a need basis) primarily focused on technical skills. -One-on-one time (weekly or bi-weekly) with PI primarily focused on big picture. -Relevant level-appropriate literature and tutorials will be provided.

Preferred Student Status

Sophomore
Junior
Return to All Project Page
Share This