physics.chem-phMar 15, 2026

Life cycle assessment for all organic chemicals

Authors: Shaohan Chen, Tim Langhorst, Julian Nöhl, Christopher Oberschelp, Martin Pillich, Johannes Schilling, André Bardow

Organizations: Laboratory of Energy and Process Systems Engineering (EPSE), ETH Zurich, 8092 Zurich, Switzerland · NCCR Catalysis, Switzerland · Chair of Ecological Systems Design (ESD), ETH Zurich, 8092 Zurich, Switzerland

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Chemicals are embedded in nearly every aspect of modern society, yet their production poses substantial sustainability concerns. Achieving a sustainable chemical industry requires detailed Life Cycle Assessment (LCA); however, current assessments face many unknowns due to limited, partly inconsistent, and untransparent data coverage since existing Life Cycle Inventory (LCI) databases account for only a tiny fraction of traded chemicals. Here, we introduce the Chemical RetrosYnthesiS for Transparent Assessment of Life-cycles (CRYSTAL) framework, which automatically generates consistent and transparent LCI data for organic chemicals based on their molecular structure using retrosynthesis and machine-learned gate-to-gate inventories. Using the predictive power of CRYSTAL, we create a consistent database for more than 70000 organic chemicals, comprising over 110000 transparent LCI datasets that quantify both feedstock and energy demands, together with associated auxiliary materials, biosphere flows, and waste flows. From this comprehensive database, we identify 50 key environmental hotspots driving high impacts of organic chemical production across multiple environmental categories and pivotal hub chemicals that are most critical for downstream chemical production. In providing this comprehensive data foundation, the CRYSTAL framework offers systematic guidance for targeted engineering and policy interventions. Its transparent, modular nature is designed to shift chemical LCA from a reliance on "unknown unknowns" to a collaboratively improvable mapping of "known unknowns".

Figures & tables

Figure 1: The CRYSTAL framework automatically and transparently predicts cradle-to-gate LCI data for organic chemical production and identifies environmentally optimal production pathways. CRYSTAL balances efficient LCI data generation with the need to assess the environmental impacts of a vast number of traded chemicals, closing critical data gaps in life cycle assessments. Based on the molecular structure of the target chemical, we 1) predict retrosynthesis pathways, 2) estimate the LCI data, i.e., material demands, energy demands, waste streams, and direct emissions for each reaction step, 3) construct a Chemical Reaction Network (CRN) to transfer information across reactions, and 4) optimize the CRN based on the Life Cycle Impact Assessment (LCIA) scores calculated from the predicted LCI data to identify environmentally optimal production pathways using user-specified environmental objectives. The framework flexibly adapts to various LCIA methods and background databases.
Figure 2: Performance of CRYSTAL on three state-of-the-art LCA databases for climate change impact. The first three columns present the validation results of the CRYSTAL framework on the ecoinvent [ 46 , 17 , 13 ] ( 257257 organic chemicals (version 3.10, cut-off) assessed with ReCiPe 2016 v1.03, midpoint (H)), cm.chemicals [ 7 , 41 ] ( 746746 organic chemicals based on IPCC 2021 global warming potential (GWP100)), and a literature database [ 26 ] ( 143143 organic chemicals assessed with ecoinvent, version 3.5, cut-off) based on ReCiPe v1.13, midpoint (H)), using the corresponding database as the basis and a Leave-One-Out approach. The “Reference Comparison” column compares the ecoinvent and cm.chemicals databases for 232232 chemicals identified as the complete intersection of the two databases with available SMILES representations, serving as a benchmark for evaluating CRYSTAL ’s performance. The violin plots of the first three columns summarize CRYSTAL ’s overall performance on each database, indicating the percentage of climate change (unit: kg\mathrm{kg} COX2\text{CO}{\vphantom{\text{X}}}_{\smash[t]{\text{2}}} -eq per kg\mathrm{kg} of the target chemical) predictions falling within an acceptable accuracy range, based on predicted-to-reference value ratios. Dashed lines indicate the acceptable range defined by the Association for the Advancement of Cost Engineering (AACE) accuracy bounds ( 50 %200 %50\text{\,}\mathrm{\%}200\text{\,}\mathrm{\%} of the environmental impact defined in the corresponding state-of-the-art database). Violin plots display the distribution between the 5th and 95th percentiles to highlight this range. The parity plots of the first three columns illustrate CRYSTAL ’s Pearson correlation and Mean Absolute Relative Error (MARE) when the last reaction step of the predicted routes is the same as the one of the reference databases.
Figure 3: Pinpointing environmental hotspots across impact categories relative to climate change to support chemical manufacturers, regulatory authorities, and LCA data providers. Top row: Climate change versus human toxicity: carcinogenic for pathways optimized for climate change impact. Chromium trioxide and anthraquinone, together with chromium-containing waste incineration, are highlighted as relative environmental hotspots in human toxicity: carcinogenic , along with their downstream chemicals (left). Regulating chromium(VI) emissions for these two hotspot reactants and for the chromium-containing waste treatment effectively reduces human toxicity: carcinogenic of all downstream chemicals (middle). Number of downstream chemicals (D/S) associated with these hotspots (right). Bottom row: Climate change versus ozone depletion for pathways optimized for (left) climate change and (middle) ozone depletion impact. Relative environmental hotspots in climate-change-optimized pathways are highlighted together with their downstream chemicals (i.e., five upstream reactants (R-hotspots) and two solvents, chloroform and dichloromethane). To emphasize changes in point density, only downstream chemicals with disproportionately high ozone depletion impacts are highlighted (middle). Number of downstream chemicals associated with these hotspots (right). ecoinvent version 3.10 (cut-off) [ 46 , 17 , 13 ] and LCIA method ReCiPe 2016 v1.03, midpoint (H) are used as the underlying LCA database.
Figure 4: Hub chemicals identified to guide prioritization of process optimization strategies for enhancing the sustainability of the chemical industry. a ) A representative example illustrating the structurally important role of a hub chemical within an illustrative chemical reaction network. The green part indicates the joint-optimal reactions across all environmental impact categories. b ) Ranking of 5252 hub chemicals by the number of strongly affected downstream chemicals, measured as those whose climate change impacts decrease by more than 2 %2\text{\,}\mathrm{\%} in response to a 10 %10\text{\,}\mathrm{\%} reduction in the impact of each hub chemical (details described in the Supplementary Text). Chemicals available in the underlying ecoinvent database are highlighted in red. c ) Switching from fossil- to bio-based ethanol pathways substantially reduces the climate change impacts for over 20002000 strongly affected downstream products.
More in this topic

Life Cycle Assessment

See all 10 papers

Explore similar work

CardsList

  1. DianShi-RxnDB: A Large-Scale, Fine-Grained Organic Reaction Data Platform Built via a Fully Automated Pipeline for Researchers and AI Agents

    Sep 6, 2026Yubin Wang, Xingjian Wei, Jiang Wu +33Document Information ExtractionAI for Science

    Shanghai Artificial Intelligence Laboratory

  2. ReTRACE: Feedback-Guided LLM Agent in a Chemistry Harness for Retrosynthesis Planning

    Oct 17, 2026Jinxiang Xie, Yijian Zhang, Qingyuan Yi +2LLM AgentsChemical Reasoning in LLMs

    State Key Laboratory of Analytical Chemistry for Life Science, Kuang Yaming Honors School, Chemistry and Biomedicine Innovation Center (ChemBIC), ChemBioMed Interdisciplinary Research Center at Nanjing University, Nanjing University, Nanjing 210023, China · School of Medicine, Guangxi University, Nanning, 530004, China

  3. Fast Organic Crystal Structure Prediction with Unit Cell Flow Matching

    Jun 2, 2026Alston Lo, Luka Mucko, Austin H. Cheng +4Flow MatchingCrystal Structure Prediction

    MIT CSAIL, Cambridge, MA, USA · University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, Croatia · Department of Chemistry, University of Toronto, Toronto, ON, Canada +8