Answers are grounded in this paper and cite their sources.
The rapid growth of the battery industry, driven by electric vehicle adoption, highlights the increasing demand for batteries and the need for sustainable, circular recycling processes to mitigate resource scarcity and environmental impacts. While numerous studies on spent battery recycling technologies exist, systematic decision-support tools remain limited. This work proposes a decision-support system comprising two complementary frameworks supported by a structured knowledge database. First, two large language models were fine-tuned using 80 named entity recognition (NER)-annotated samples and 150 question-answer (Q&A) pairs derived from over 10k+ articles. The NER model enabled the construction of a battery recycling-informed knowledge graph (BR-KG), which was integrated with a Q&A model, the Battery-LLaMA module. Second, a multi-objective screening (MOS) framework was developed to rank recycling technologies based on revenue generation, energy consumption, and recycling carbon footprint. To evaluate performance, 30 technical Q&A pairs focused on material efficiency and processes were curated, and Battery-LLaMA demonstrated improved performance compared to ChatGPT-4o (F1-score: 0.821 vs. 0.701), highlighting enhanced quantitative and qualitative responses. The MOS framework, implemented using a weighted-sum approach, revealed trade-offs between economic and environmental metrics. Among the top eight-ranked processes, low-temperature leaching showed the lowest recycling carbon footprint (122.84 kg CO 2 − e q ) but relatively high energy consumption (351 MJ), with a revenue of $769.61, while other preferable solutions are summarized in Table 3. Overall, this work demonstrates how AI-driven frameworks can guide the battery industry toward a more sustainable and circular economy by identifying and implementing optimal battery recycling pathways.
No takes yet. Share an insight, caveat, or question.
Garampalli et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: