×

Machine Learning Approach for CO2 Reduction through Sustainable Chemical Transformations

Professor: Prof. Debabrata Maiti

Description:

Empowering a sustainable future with innovative models, enhance reaction forecasting, and predict the impacts of reaction parameters on solving energy problems and catalyse sustainable development. In particular, we will focus on developing critical applications of machine learning in industrial energy issues and sustainable development. Cutting edge research with hand-on experience and data analysis.

Objectives:

The goal of this project is to utilize machine learning techniques to accelerate and optimize chemical transformations that reduce CO2 emissions or convert CO2 into valuable products. This project will focus on the following objectives:

1. Catalytic CO2 Reduction and Utilization: Apply machine learning to discover and optimize catalysts that facilitate the conversion of CO2 into valuable chemicals (e.g., fuels, polymers, or industrial feedstocks), reducing greenhouse gas levels.


2. Sustainable Reaction Pathways for CO2 Conversion: Use data-driven models to identify sustainable synthetic pathways for CO2 fixation and conversion, maximizing atom economy and minimizing environmental impact.


3. Optimizing Reaction Conditions for CO2 Fixation: Develop machine learning models to predict optimal reaction conditions (e.g., temperature, pressure, solvent, catalyst) for CO2 reduction reactions, promoting greener and more efficient processes.


4. Green Chemistry and Energy Efficiency: Implement machine learning techniques to design energy-efficient reactions that reduce the energy demand in CO2 conversion processes, adhering to green chemistry principles.

Methods:

The project will employ a combination of machine learning algorithms and organic synthesis expertise to drive sustainable CO2 reduction and transformation. Below are the details:

Data Collection and Preprocessing:

• Collect experimental data from our lab’s past research on CO2 conversion reactions and from open-source chemistry databases (e.g., Reaxys, PubChem).


• Encode molecular structures using SMILES notation and molecular fingerprints, and compile reaction data (catalysts, solvents, conditions, yields).

Machine Learning Model Development:

• Catalyst Prediction Models: Train supervised learning models (Random Forest, XGBoost, Neural Networks) to predict the performance of various catalysts in CO2 conversion reactions, using molecular descriptors and reaction conditions as inputs.


• Reaction Condition Optimization: Apply reinforcement learning and optimization algorithms to identify optimal reaction conditions (temperature, pressure, solvent, time) that maximize CO2 reduction efficiency while minimizing waste.


• Retrosynthetic Analysis for CO2 Utilization: Develop machine learning models to suggest alternative, sustainable synthetic routes that use CO2 as a feedstock, optimizing for atom economy and environmental friendliness.

Data Collection and Preprocessing:

• Collect experimental data from our lab’s past research on CO2 conversion reactions and from open-source chemistry databases (e.g., Reaxys, PubChem).


• Encode molecular structures using SMILES notation and molecular fingerprints, and compile reaction data (catalysts, solvents, conditions, yields).

Model Validation and Experimental Feedback:

• Perform cross-validation to evaluate model performance (R², Mean Absolute Error, etc.) and refine models based on experimental feedback from CO2 conversion reactions carried out in the lab.


• Perform chemical reactions to verify machine learning predictions and optimize reaction conditions in real-world lab settings.

Sustainable Lab Integration:

• Develop an interactive tool for researchers to input experimental data and receive machine learning-driven recommendations for CO2 reduction and transformation processes.

Deliverables:

The project will generate several impactful outcomes:

1. Predictive Models for CO2 Reduction: Accurate machine learning models that predict the most efficient catalysts and reaction conditions for converting CO2 into valuable products (e.g., fuels, chemicals).


2. Optimized Reaction Conditions for Sustainable CO2 Fixation: A dataset of optimized reaction conditions that adhere to green chemistry principles, reducing energy consumption and environmental impact in CO2 reduction processes.


3. Sustainable Reaction Pathways for CO2 Utilization: A curated list of alternative reaction pathways using CO2 as a raw material for creating industrial feedstocks, ensuring the best balance between efficiency, atom economy, and environmental sustainability.


4. Open-Source Tools and Educational Materials: An open-source machine learning tool that allows other researchers to input reaction conditions and molecular structures to receive optimized recommendations for CO2 reduction reactions. Educational materials (e.g., case studies, tutorials) that will simplify the technical aspects of machine learning for undergraduate students interested in sustainability research.

Link to Sustainability:

This project is directly aligned with global sustainability efforts to reduce CO2 emissions and combat climate change. The deliverables from this project address several UN Sustainable Development Goals (SDGs):

• SDG 13 (Climate Action): By focusing on the reduction of CO2 emissions and its conversion into valuable products, this project directly contributes to climate action by mitigating the primary cause of global warming.


• SDG 7 (Affordable and Clean Energy): By designing energy-efficient reactions and optimizing the use of renewable feedstocks, this project supports the development of cleaner, more affordable energy solutions.


• SDG 9 (Industry, Innovation, and Infrastructure): The application of machine learning to organic synthesis for CO2 reduction fosters innovation in the chemical industry, paving the way for sustainable industrial processes.

Conclusion:

This project bridges the gap between machine learning and sustainable organic synthesis by focusing on the critical issue of CO2 reduction. Through data-driven predictions and optimizations, we aim to significantly enhance the efficiency of chemical processes that utilize CO2, reducing emissions and contributing to the global effort to combat climate change.

The proposed research will provide undergraduate students involved in the SPURS program with hands-on experience in using advanced data science tools in a chemistry lab setting. By applying machine learning to real-world problems like CO2 reduction, this project will not only foster sustainability but also equip students with valuable interdisciplinary skills.

Pre-requisites:

• Mathematics (Linear Algebra, Calculus, Statistics), Computer Programming (Python or Java, Data structure and Algorithm)

×

Climate damages associated with food trade in India

Professor: Prof. Srinidhi Balasubramanian

Description:

Food systems emit a third of the greenhouse gases yet are unregulated and unaccounted in many climate action plans. Greener diets have also been linked to better health outcomes. The trilemma of diets, health and climate change is exacerbated in a country like India that has large inequalities in access to food. One key driver to the delivery of food commodities is trade. In this project, the student will adopt an Environmentally Extended Input-Output Approach (EEIO) to explore how trade affects both nutritional and climate outcomes. The work will include a systematic literature survey on trade and climate change, collate data from expenditure surveys, emission databases, MIRO assessments, trade portals, and identify heterogeneity in nutritional outcomes and commodity specific climate footprints as mediated by international trade.

Objectives:

1. Estimate the environmental impacts of food trade with focus on land use, energy use and embodied greenhouse gas emissions


2. Track virtual nitrogen flows stemming from food trade

Methods:

• Track the evolution of imports and exports of food commodities in India


• Identify food commodity-specific emission factors associated with emission factors in India


• Develop an input-output model using agriculture, trade, and land use/land cover databases to track resource inputs and emission outputs from the Indian food system.

Deliverables

• Primary goal is to develop a greenhouse gas emissions inventory and a reactive nitrogen budget from sectors driving food trade in India. Support ongoing efforts in developing publications with colleagues in the CARES group.

Pre-requisites:

• Preferably has cleared ES200 or equivalent introductory Environmental Science and Engineering course(s).

×

Assessment of water sustainability for green hydrogen production

Professor: Prof. Yogendra Shastri

Description:

Green hydrogen is being promoted as an alternative fuel to reduce carbon emissions related to energy consumption. Green hydrogen can be used as fuel in combustion or as well as in fuel cells for transport applications. Additionally, it is an important feedstock in chemical industries. Green hydrogen is produced by water electrolysis (water splitting), and large quantities of water are needed to produce hydrogen.

Objectives:

The goal of this project will be to perform a detailed sustainability assessment of green hydrogen production from a water standpoint. Specific objectives are as follows:

1. Quantify the water footprint of different green hydrogen technologies


2. Assess future water footprint considering expected green hydrogen demand


3. Map water footprint with water availability


4. Provide recommendations on preferred technologies and regions

Objectives:

1. Calculation of Solar and Wind Energy availability over India for last 10 years using ERA5 data


2. Identification of weather regimes in Solar and Wind Energy availability using traditional EoF and data driven K-Means Clustering algorithms


3. Comparing the regimes with the well-known active-break cycles of monsoons

Deliverables:

The work cuts across the three fields of energy, sustainability, and climate. The student will work on developing simple models to estimate water footprint. The specific deliverables are:

1. Detailed quantification of water footprint for different green hydrogen technologies


2. Preliminary recommendations for promising technologies

The work will be completely computational in nature.

Pre-requisites:

• There is no prerequisite as such. The student should have a basic knowledge of material and energy balances, and should be interested in working on a computational problem.

×

Biochar and ERW as viable CDR technologies in an Indian context

Professor: Prof. Vikram Vishal

Description:

Carbon Dioxide Removal (CDR) encompasses various strategies aimed at removing CO2 from the atmosphere to combat climate change. The global market for CDR technologies is expected to grow significantly, potentially reaching $4 billion by 2030. Among CDR methods, biochar and enhanced rock weathering (ERW) are gaining traction due to their scalability, stable sequestration, and tremendous potential for addressing both climate change and ecological sustainability.

Biochar is a stable form of carbon produced from biomass through pyrolysis, which can sequester carbon in the soil, while ERW involves the application of crushed silicate rocks to enhance natural weathering processes that absorb CO2 from the atmosphere.

Objective:

1. To deliver actionable insights into the region-specific suitability, economic feasibility, carbon sequestration potential, and co-benefits of Biochar and ERW for businesses looking to invest in CDR technologies. We aim to decipher which method offers better returns on investment (ROI) in terms of scientific robustness, carbon credits, operational costs, and environmental co-benefits.

Deliverables:

1. Scientific Robustness and Technology Readiness Analysis: Examining the reliability of aforementioned protocols via literature review and project reports to identify gaps and compliance with Monitoring, Reporting and Verification protocols.


2. Economic Feasibility Study: A comparative analysis showing the cost per ton of CO2 removed for both Biochar and ERW, alongside projected returns from carbon credits or offset markets


3. Life Cycle Assessment (LCA): A detailed LCA with upstream and downstream emissions analysis encompassing required infrastructure, raw materials (biomass for Biochar, silicate minerals for ERW), and regional suitability for each technology.


4. Co-benefits and Market Differentiation Analysis:


• A detailed account of the co-benefits such as improved soil fertility, waste reduction, and potential synergies with existing operations.


• Business-specific strategies for leveraging these co-benefits to enhance sustainability profiles or reduce operational costs.


5. Risk and Regulation Strategy Brief:


• A risk assessment highlighting potential regulatory, logistical, and environmental challenges.


• Policy recommendations for businesses to navigate evolving carbon markets, including potential government incentives or tax benefits for deploying Biochar or ERW technologies