LLM-Based Data Extraction and Machine Learning for CO2e Estimation of Semiconductor Components

Abstract

This thesis investigates the application of Artificial Intelligence (AI) and Machine Learning (ML) methods to improve carbon footprint estimation for Integrated Circuit (IC) components in automotive Cost Engineering (CE). The work addresses challenges faced by electrical cost engineers at Volvo Cars: the manual, time-consuming process of collecting component specifications, and estimating Carbon Dioxide Equivalent (CO2e) emissions from electronic components.

We developed an automated data collection pipeline using Large Language Models (LLMs) to extract structured information from manufacturer datasheets and material content sheet documents. Three models, Gemma 3 4B, Llama 3.1 8B, and gpt-oss 120B, were evaluated for extraction accuracy, inference time and memory usage. The gpt-oss 120B model achieved 98.5\% extraction accuracy for the validation set. The data extraction pipeline was then applied to a larger set of datasheets and material content sheets and converted unstructured PDF documents into structured tabular data to be used for machine learning.

We developed machine learning models to predict the masses of four hotspot metals, copper, gold, palladium and silver, which are the primary contributors to raw-material carbon emissions. We evaluated TabPFN (a transformer-based prior-fitted network), XGBoost and CatBoost regression models. CatBoost achieved the best overall performance, with R2 values of 0.994 for gold and 0.971 for palladium. SHAP analysis revealed that total component mass is the most important feature for copper content, while pin count has the biggest effect on gold and silver predictions.
We introduced a simplified hotspot scaling approach that reduces the previous model's complexity from ten function-type-specific scaling factors to a single global factor, with minimal loss in accuracy.


The model was integrated into a desktop application that enables cost engineers to retrieve component specifications, predict metal content when material data are unavailable, and compute CO2e emissions automatically. The tool reduces manual effort and provides a reproducible, consistent framework for carbon footprint estimation in component evaluation.

This work demonstrates that AI and ML methods can effectively automate and improve semiconductor carbon footprint estimation, supporting Volvo Cars' decarbonisation objectives while reducing the manual workload of cost engineering teams.

Popular Abstract

Modern cars contain hundreds of electronic components, tiny chips that control everything from the engine to the entertainment system. But these small components come with an environmental cost, the carbon dioxide released during their production. For a company like Volvo Cars, which aims to eliminate all greenhouse gas emissions by 2040, understanding and reducing the carbon footprint of every single component is crucial.

The problem is that tracking these emissions is incredibly time-consuming. Engineers must manually search through technical documents from manufacturers, extract information about each component's size, weight and material composition, and then calculate the associated carbon emissions.

This thesis tackles that challenge by using machine learning to make that process easier. We developed a system that uses artificial intelligence to automatically read and extract information from manufacturer documentation. The AI can scan through technical PDFs, find the relevant information (such as how much gold, silver, or copper a component contains), and organize it into a structured format.

However, manufacturers don't always publish complete material information, so we also developed machine learning models that can predict a component's metal content based on characteristics like its size, weight and number of pins. The models proved accurate, particularly for gold, the metal that contributes most to carbon emissions from raw-materials in electrical components.

All of this technology was packaged into a user-friendly desktop application that engineers at Volvo Cars can use in their daily work. By entering a component's identification number, the tool can retrieve specifications, predict metal content if needed, and calculate the estimated carbon footprint. What once took an excessive amount of time and manual labor, now takes seconds, and what once required expert knowledge is now accessible to any engineer.

• author: Müller, Arvid and Flynn Rosenberg, Elias
• supervisor: Amir Aminifar
• organization: Department of Electrical and Information Technology
• course: EITM01 20261
• year: 2026
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• keywords: large language models, machine learning, carbon emission, semiconductor, IC, data extraction, tabular data, gradient boosting, co2e, cost engineering, automotive industry, shap, xai, sustainability, life cycle assessment, information extraction
• report number: LU/LTH-EIT 2026-1160
• language: English
• id: 9239103
• date added to LUP: 2026-06-17 10:16:26
• date last changed: 2026-06-17 10:16:26