The stakes are higher than they might appear. Life cycle assessment, or LCA, is the standardised method for quantifying the potential environmental impacts of products and services from raw material extraction through manufacturing, use, and disposal. Yet the method has long been constrained by limited access to primary data, meaning company-specific, measurement-based information. Most practitioners instead rely on secondary life cycle inventory datasets that often lack geographical, temporal, and technological representativeness. Previous studies have documented the consequences: assessments of electric mobility are hampered by scarce primary data for battery raw material supply chains, and researchers have repeatedly flagged persistent challenges in traceability and interoperability across inventory databases. The field has recurrently called for better access to primary data, along with standardised metadata covering temporal representativeness, location, technology specifications, and energy sources.

This is where digital product passports enter the picture. Originally conceived to support supply chain traceability, regulatory compliance, and circular economy objectives, passports could in principle reduce reliance on outdated secondary datasets by embedding harmonised, machine-readable, product-level information on material composition, product characteristics, and environmental indicators. The relationship is potentially bidirectional: while passports improve access to verifiable product-level data, LCA offers a science-based, standardised methodological framework for structuring and interpreting that information, including established principles for defining system boundaries and assessing data quality. The Leiden-led team set out to move beyond conceptual wish lists and ask how real, emerging passport initiatives actually perform in practice.

The 20 initiatives examined span batteries, electronics, construction products, textiles, and other material-intensive sectors, and include research projects, policy platforms, industrial implementations, and standardisation efforts. The sample was assembled purposively from publicly documented initiatives rather than as a systematic or exhaustive inventory, and the researchers grouped the LCA-relevant attributes they found into four categories: assessment results, product characteristics, production characteristics, and end-of-life characteristics. The good news is that several core attributes are now common across initiatives. Material composition, sometimes reported with absolute masses, recycled content, hazardous substances information, and product carbon footprints appear frequently and could provide a genuine basis for improved LCA practice. Notably, many of these attributes require limited incremental effort, because companies already collect them for compliance with regulations such as the restriction of hazardous substances directive and the REACH chemicals regulation.

The dominant logic across many initiatives is the exchange of aggregated product carbon footprints along the value chain. Frameworks such as the World Business Council for Sustainable Development’s Pathfinder and PACT, sectoral standards like Together for Sustainability in the chemical industry, and the automotive data space Catena-X all follow a similar inheritance-based calculation model: each supply chain actor calculates a cradle-to-gate carbon footprint based on its own processes, passes the aggregated result downstream, and the next user multiplies the received figure by the quantity of input product to derive its own footprint. This approach reduces the reporting burden considerably, but it creates comparability challenges. Consistent use of carbon footprints requires alignment in methodological and modelling choices, such as system boundaries and background inventory datasets, which is difficult to guarantee in heterogeneous supply chains. Although sectoral guidelines like the Catena-X carbon footprint rulebook and the Global Battery Alliance calculation framework attempt harmonisation, the researchers found that the analysed initiatives prioritise technical interoperability and data exchange formats over full end-to-end methodological alignment.

The deeper problem is granularity. Aggregated carbon footprints mask the processes, technological parameters, geographical conditions, and operational behaviours that drive environmental impacts. For most LCA-based decisions, such aggregation is overly simplified: researchers need process-specific data to identify impact hotspots across a product life cycle, assess whether impacts are merely shifted elsewhere, or design mitigation strategies. Yet generating and sharing process-specific inventory data is genuinely difficult in practice. It increases reporting burden, requires frequent updates, and may expose commercially sensitive information such as production efficiencies and technology choices. The trade-off, the authors argue, is not only technical but organisational and governance-related: the data that is most informative for LCA is often also the least shareable within supply chains. Some passport architectures already include tiered access rights to manage this tension, restricting detailed data to accredited actors such as regulators or auditors, which strengthens verification but may limit broader analytical use.

The asymmetry extends across life cycle stages. The analysed initiatives predominantly capture upstream and mid-stream product attributes, while providing limited coverage of production conditions, use-stage performance, and end-of-life outcomes. Factory energy use and sources, geographical and temporal metadata, and end-of-life parameters appear far less frequently than composition data. The researchers emphasise that this is not simply a design flaw but reflects structural differences in data controllability. Upstream and mid-stream information sits closer to the manufacturer’s sphere of influence and is often already collected for procurement, compliance, traceability, or quality control. In contrast, use and end-of-life outcomes depend on factors beyond direct producer control, including user behaviour, local infrastructure, and downstream waste management systems. As a result, downstream information in current passports mainly reflects intended use conditions, repair guidance, or recyclability claims rather than measured outcomes, marking a structural boundary to what passports can realistically provide at the point of product placement.

End-of-life modelling deserves particular attention. In LCA practice, end-of-life modelling combines factual data, such as recycling or landfilling rates, with scenario assumptions. If end-of-life impacts are embedded within an aggregated carbon footprint, they must be sufficiently disaggregated to separate assumed downstream scenarios from upstream processes. But dynamic, product-specific end-of-life performance data is rarely available in verifiable form and generally falls outside the practical scope of passport systems. Downstream actors are unlikely to feed operational data into individual passports, and many end-of-life outcomes are inherently modelled rather than measured. The authors caution that expecting passports to resolve these limitations risks conflating product-level information systems with broader waste-system monitoring infrastructures. Instead, they argue, passports should frame end-of-life information as scenario input: structured guidance on plausible product-specific pathways that supports more consistent downstream assumptions without exceeding industry’s reporting capacity.

The way forward, according to the team, involves embedding established LCA principles directly into passport data models. That means integrating system boundaries, functional units, provenance rules, and data quality indicators, so that passport information remains interpretable and comparable across supply chains. Transparency in current LCA practice is itself uneven, with many datasets in widely used inventory databases lacking accessible documentation, consistent metadata, or traceable sources. Passports could help by requiring product-level data in machine-readable, interoperable formats with explicit provenance and methodological metadata, and by clearly labelling proxy data that can be replaced when verified primary information becomes available. Verification itself should be understood in a differentiated sense, ranging from self-declared and documented data to audited or measured values. Passports cannot eliminate diversity in LCA practice, the authors note, but they can make that diversity visible and auditable, turning opaque uncertainty into an attribute that can be examined and managed. Dynamic update mechanisms, such as change logs and timestamps, could further allow passports to function as living datasets rather than static records.

Ultimately, the researchers conclude, digital product passports and life cycle assessment can be mutually reinforcing: passports offer a pathway to structured, product-level information, while LCA provides the methodological backbone for interpreting it. Realising that complementarity requires deliberate shifts in design and governance, including moving beyond aggregated indicators towards disaggregated inventory data on energy use, critical raw material intensity, and provenance, while recognising that passports should complement rather than replace existing inventory databases. Harmonised reporting conventions aligned with ISO and Product Environmental Footprint principles would make assessments more representative, transparent, and reproducible. If those shifts materialise, the humble product passport could evolve from a static compliance tool into a dynamic infrastructure for actionable sustainability assessment, one that links product information with material, environmental, and socio-economic flows for regulators, auditors, repairers, recyclers, and consumers alike.

Subject of Research: Data structuring in digital product passport initiatives and their support for life cycle assessment

Article Title: Emerging digital product passport initiatives require improved data structuring to support life cycle assessment

Article References: Liang, Y., Mogollón, J. M., Cunha, S., Donati, F., Li, C., Mintjes, B., Kleijn, R., Climent, D. M., Wang, R., Muller, S., Mas-Fons, A., Xicotencatl, B. M., & Istrate, R. (2026). Emerging digital product passport initiatives require improved data structuring to support life cycle assessment. Journal of Industrial Ecology, 30(4), 2131-2139. https://doi.org/10.1007/s44498-026-00144-7

Image Credits: AI Generated

DOI: 10.1007/s44498-026-00144-7

Keywords: digital product passport, life cycle assessment, life cycle inventory, product carbon footprint, circular economy, ecodesign regulation, end-of-life modelling, data quality, supply chain traceability, industrial ecology, sustainability data, EU regulation

The concept of weighing the hidden material burden of products dates back to the early 1990s, when Friedrich Schmidt-Bleek introduced the MIPS method, Material Input Per Service unit, famously popularized as the ecological backpack. The metaphor captures the idea that every object carries a load of extracted natural resources far beyond the materials it visibly contains. Over the decades this idea matured into the material footprint, a member of the footprint family alongside the carbon and ecological footprints. At the national scale, material footprints are estimated through input-output analysis and attribute all globally mobilized resources to final consumers. At the product scale, the equivalent indicator is the Product Material Footprint, or PMF, typically calculated within a Life Cycle Assessment, the ISO 14040-standardized framework that inventories all inputs and outputs of a product system and then translates them into environmental impacts.

Herein lies a technical puzzle. In standard life cycle impact assessment, elementary flows drawn from the inventory are converted into impact scores using characterization factors, precomputed multipliers that link each flow to a normalized effect. For the material footprint, researchers at the Wuppertal Institute and later Mostert and Bringezu developed sets of characterization factors mapped onto ecoinvent’s resource flows, and the most recent update by Mostert and colleagues in 2025 represents the most comprehensive PMF method available. But this mapping approach has a weakness: characterization factors are typically aggregated globally, often as medians, and matching them to inventory flows introduces unknown errors, particularly for metal ores that are regionally underrepresented or absent. The Wuppertal team proposes an alternative they call direct PMF calculation, or D-PMF, which reads the resource extraction flows already documented in ecoinvent’s life cycle inventories and sums them with characterization factors of one, essentially trusting the database’s own bookkeeping rather than an external overlay.

The PMF consists of two impact categories with distinct meanings. Raw Material Input, or RMI, accounts for used extractions from the environment: all raw materials sold by mining, agriculture, forestry and fisheries that enter a product over its complete life cycle. Total Material Requirement, or TMR, goes further and includes unused extraction, the natural material that must be moved and dumped to enable extraction, such as overburden in coal mining or crop residues left in fields. For abiotic resources, the researchers exploited ecoinvent’s convention of recording metal and mineral extraction as the target material plus a separate gangue flow, so that the total extracted mass is captured. For the unused part, they created a new elementary flow called overburden, attached to the waste flows that ecoinvent uses for mining spoils. For biotic resources, where ecoinvent rarely documents biomass as an elementary flow, the team built on an approach by Stefan Pauliuk and used the gross calorific value of biogenic products, dividing the documented energy content by an average energy density of 19.5 megajoules per kilogram to estimate dry biomass. Unused biomass was estimated from residue-to-crop ratios found in the literature, with water content corrections for agricultural products.

The method was implemented as a life cycle impact assessment method in the open-source Python framework Brightway2, and the scripts were released on GitHub. The team then compared their direct calculation against the characterization-factor benchmark across a broad product set: fifteen major industrial metals, ten common minerals, five fossil fuels, ten agricultural products, five wood products, and three everyday exemplars, a laptop, a building and one kilowatt-hour of German electricity. The results reveal a split verdict. For most minerals and energy carriers, the two methods agree within a factor of two. For metals, however, deviations can reach an order of magnitude, and the reasons are instructive. Characterization factors are global medians, while ecoinvent datasets are often regionalized. The database’s tin dataset, for example, reflects an underground mine in Peru, whereas most tin is mined above ground, producing less gangue than the global average assumes. Gold is even more extreme: gangue values in ecoinvent version 3.11 range from roughly 23,000 to over two million kilograms per kilogram of gold, a variance that no single global factor can capture.

The product-level consequences are striking. For the laptop, a metal-intensive product whose footprint is dominated by gold production in integrated circuits and printed wiring boards, the direct method yields an RMI 27 percent lower than the benchmark, 531 kilograms instead of 725, and a TMR 53 percent lower, 823 kilograms instead of 1,775. At the resource level, the benchmark method’s largest contributor is the flow for gold itself, while the direct method’s largest contributors are gangue and overburden, exposing how differently the two approaches allocate the hidden mass of mining. For the building, which draws mainly on granite, gravel and clay, deviations shrink to between 0.5 and 12 percent. Electricity tells the most dramatic TMR story: while RMI values are close, 0.53 versus 0.59 kilograms per kilowatt-hour, the direct method’s TMR is five times higher, 3.5 versus 0.71 kilograms, driven by overburden in lignite and hard coal mining. Notably, the direct values for coal are corroborated by independent German federal data suggesting 9 to 13.5 kilograms of overburden per kilogram of lignite, lending credibility to the database figures.

The deeper finding concerns data coverage. Analyzing ecoinvent versions 3.4 through 3.11, the researchers found that the share of mining and metallurgical processes containing the gangue flows they should contain has climbed to 99 percent, a highly beneficial trend that makes direct RMI calculation genuinely viable. Overburden coverage, by contrast, has stagnated or even declined, sitting at roughly 50 percent of relevant processes. This gap makes direct TMR quantification impossible without supplementing the database with external data, which undermines the very consistency and transparency advantages that motivated the direct approach. The same limitation applies to biotic TMR, since ecoinvent does not monitor unused biomass at all, and the residue-to-crop ratios used to estimate it carry high uncertainty. Even the biotic RMI workaround has quirks: using an average energy density overestimates extraction for high-energy products such as cork or oil seeds, while the tomato case showed the energy-based estimate of 0.05 kilograms of dry biomass per kilogram integrating well with the fruit’s known 92 percent water content.

Beyond the numbers, the study articulates an epistemological tension within industrial ecology. Material flow accounting treats absolute physical mass extraction from nature as an environmental pressure in its own right, whereas utility-focused impact assessment methods frame resource issues around depletion risks, supply constraints or downstream damages. The direct PMF method operationalizes the former philosophy within a bottom-up, process-based framework, aligning conceptually with input-side indicators like cumulative energy demand. It cannot eliminate allocation challenges for multi-metal ores, where economic allocation based on volatile commodity prices remains necessary, but by adopting ecoinvent’s own allocation rules it minimizes the mixing of data sources and applies uniformly across all ecoinvent system models, including Cut-off, Consequential and APOS. It also preserves geographic specificity that global factors erase, although the authors caution that a single highly specific dataset, like the Peruvian tin mine, can mislead when used to model generic global production.

The practical implications reach into corporate procurement, product design and sustainability reporting, especially as the European Union’s Ecodesign for Sustainable Products Regulation introduces mandatory digital product passports that could improve supply chain data availability. The authors argue that a coordinated harmonization initiative among public authorities, data providers and scientific organizations is needed to make TMR directly calculable, noting that most detailed mining datasets are currently commercial and expensive. Their conclusion is candid: direct calculation from ecoinvent is a valid option for Raw Material Input today, but not yet for Total Material Requirement. The quantification of the product material footprint, they observe, is ultimately not a methodological problem but one of data collection, and database providers are urged to document resource extraction with the same rigor now applied to primary energy content and carbon dioxide emissions. Future studies, they suggest, would benefit from reporting both methods side by side, particularly where metal ores and overburden dominate the results.

Subject of Research: Direct calculation of product material footprints from life cycle inventory data in the ecoinvent database

Article Title: Re-evaluating ecoinvent’s suitability for direct product material footprint calculation

Article References: Buschbeck, C., Bickel, M., Teubler, J., & Liedtke, C. (2026). Re-evaluating ecoinvent’s suitability for direct product material footprint calculation. Journal of Industrial Ecology. https://doi.org/10.1007/s44498-026-00183-0

Image Credits: AI Generated

DOI: 10.1007/s44498-026-00183-0

Keywords: material footprint, ecoinvent, life cycle assessment, raw material input, total material requirement, resource extraction, industrial ecology, mining, overburden, sustainable development goals, life cycle inventory, Wuppertal Institute

The research, led by Zirui Tang and Qingshi Tu of the Sustainable Bioeconomy Research Group at the University of British Columbia, together with Max Dreger and Kourosh Malek of Forschungszentrum Jülich in Germany and Peijin Jiang of UBC, addresses a bottleneck that has long frustrated the sustainability field. Life cycle assessment, or LCA, is the standard tool for comparing the environmental credentials of products and services, but its reliability hinges on high-quality inventory data. Collecting that data by hand from the literature is slow, labor-intensive, and notoriously difficult to scale, which means that meta-analyses and large-scale syntheses often lag years behind the primary studies they should be aggregating.

The team’s framework combines two complementary technologies. On the front end, a retrieval-augmented generation pipeline—commonly known as RAG—sends large language models into the full text of published articles to extract three core categories of information: life cycle inventory data, life cycle impact assessment results, and the modeling assumptions that underpin each study. On the back end, the extracted information is normalized and mapped into an ontology-driven knowledge graph built specifically for LCA data and implemented in the Neo4j graph database. The ontology gives every extracted fact a defined place in a shared semantic structure, turning a pile of heterogeneous paper-specific numbers into an interconnected, machine-queryable web of knowledge.

The technical challenge is considerable. Scientific papers do not present their data uniformly; inventory tables appear in different formats, use inconsistent units and system boundaries, and embed crucial context—such as functional units, allocation choices, and geographic scope—in prose rather than structured fields. The pipeline therefore has to parse documents, identify which tables actually contain inventory data, and interpret the surrounding text to understand what each number means. The authors report that the extraction stage achieved F1-scores ranging from 75.18 to 89.50 percent across the three data types, a level of semantic accuracy that suggests the models are genuinely understanding the content rather than merely pattern-matching on keywords.

Extraction alone, however, would only produce a better-organized archive. The second half of the framework is what makes it genuinely interactive: an LLM-based question-answering system that translates natural language queries into executable graph queries. A user can ask, in ordinary English, about the greenhouse gas intensity of a particular bio-based chemical under a given set of assumptions, and the system converts the question into Cypher—the query language of Neo4j—retrieves the relevant nodes and relationships from the knowledge graph, and returns an answer. Crucially, users need no prior knowledge of the graph’s schema, which has historically been one of the biggest barriers preventing domain scientists from exploiting structured databases directly.

The query system itself was evaluated rigorously, and the results reveal an instructive lesson about how to build such tools. A baseline approach relying on a single strategy achieved an F1-score of only 56.98 percent—hardly reliable enough for scientific work. By combining similarity search, which locates the most relevant portions of the graph for a given question, with text-to-Cypher reasoning, which generates the formal query, the team raised the score to 75.18 percent. The hybrid architecture outperforms either method alone because the two strategies fail in different ways: semantic search is robust to vague phrasing but imprecise about graph structure, while query generation is precise but brittle when a question does not map cleanly onto the schema.

To demonstrate the framework in practice, the researchers applied it to a case study of LCA studies on bio-based chemical production—a domain where the literature has expanded rapidly as industry and academia race to replace fossil-derived chemicals with renewable alternatives. Bio-based chemicals are an ideal testbed because published assessments vary widely in feedstock, conversion technology, and impact categories, making manual synthesis across studies especially painful. The knowledge graph lets researchers compare results across studies while preserving the assumptions that make each figure meaningful, something a simple spreadsheet of extracted numbers could never do.

The broader significance extends beyond one application domain. The study arrives amid a wave of efforts to apply large language models to scientific data extraction, from polymer properties to carbon capture technologies to toxicology reports. What distinguishes this work is the full pipeline: it does not stop at extraction but carries the data through normalization, ontological mapping, and interactive retrieval, addressing the interoperability problem that has long plagued LCA data systems. Previous semantic efforts in the field, including minimal ontology patterns for LCA data proposed over the past decade, laid conceptual groundwork, but the manual effort required to populate such structures remained prohibitive. Automating that population step changes the economics of knowledge synthesis entirely.

There are, of course, caveats. F1-scores in the 75 to 90 percent range mean that a meaningful fraction of extracted facts will still be wrong or missed, and in a field where numbers feed policy decisions and corporate sustainability claims, errors carry consequences. The authors make their ground-truth datasets, extracted results, and pipeline code publicly available through the article’s supplementary materials and a GitHub repository, which allows the community to audit performance and build improvements. The work was supported by the Natural Sciences and Engineering Research Council of Canada, and the authors declare no competing interests.

Still, the trajectory is clear. As the volume of scientific literature grows faster than any human team can read, the bottleneck in fields like industrial ecology is shifting from generating data to finding and reconciling it. A framework that reads the papers, structures the findings, and answers questions in plain language offers a glimpse of how sustainability science might scale its evidence base to meet the pace of the energy and materials transition—turning decades of scattered assessments into a single, queryable map of environmental knowledge.

Subject of Research: Automated extraction and retrieval of life cycle assessment data from scientific literature using large language models and knowledge graphs

Article Title: From literature to knowledge graphs: automated extraction and retrieval of life cycle assessment data with large language models

Article References: Tang, Z., Dreger, M., Jiang, P., Malek, K., & Tu, Q. (2026). From literature to knowledge graphs: automated extraction and retrieval of life cycle assessment data with large language models. Journal of Industrial Ecology, 30(4), 1985-2000. https://doi.org/10.1007/s44498-026-00135-8

Image Credits: AI Generated

DOI: 10.1007/s44498-026-00135-8

Keywords: life cycle assessment, large language models, knowledge graphs, retrieval-augmented generation, life cycle inventory, data extraction, Neo4j, text-to-Cypher, sustainability, industrial ecology, natural language processing, bio-based chemicals

The scale of the release is striking. At its core, the database contains 2,000 unit process datasets with high spatial resolution, spanning a wide range of industry sectors. Each unit process is the fundamental building block of a life cycle inventory: a discrete industrial operation, such as producing a tonne of clinker, smelting a batch of aluminum, or generating a megawatt-hour of electricity, described in terms of its inputs of raw materials and energy and its outputs of products, emissions, and waste. By documenting these processes with detailed records of material use, energy consumption, and waste generation, TianGong gives analysts the raw ingredients from which complete product life cycles can be assembled and audited.

Beyond the unit processes, the database incorporates 157 input-output datasets and 2,277 environmental reference parameter datasets. The input-output component is particularly significant for macro-scale analysis. Input-output datasets capture the flows of goods and money between economic sectors, allowing researchers to trace environmental pressures through entire supply networks and to estimate impacts where bottom-up process data are incomplete. The environmental reference parameters provide the contextual values, such as emission factors and regional characteristics, against which specific processes can be normalized and validated. Together, these three layers form a hybrid structure that supports both fine-grained product assessments and economy-wide analyses of China’s industrial metabolism.

What distinguishes TianGong from many previous efforts is its commitment to openness, traceability, and transparency. The database follows internationally recognized methodologies and is formatted for the ILCD entry-level standard, the reference format developed under the International Reference Life Cycle Data System for exchanging life cycle inventory data. This means datasets are not merely numbers in a spreadsheet; they carry structured metadata describing their sources, system boundaries, geographic and temporal scope, and data quality. The entire database is freely accessible under the MIT license, one of the most permissive open-source licenses available, allowing researchers, companies, and policymakers to use, adapt, and redistribute the data with minimal restriction.

The problem the database addresses is well documented in the methodological literature. Studies of spatial scale effects have shown that life cycle inventory results can shift substantially depending on where and how underlying data were collected, and assessments of Chinese products built on generic global datasets can misstate impacts by wide margins. China’s industrial system has distinctive features: a coal-heavy electricity grid whose carbon intensity varies dramatically by province, process technologies that differ from those in Europe or North America, and supply chains that are often more vertically integrated than their Western counterparts. When analysts substitute foreign data for missing Chinese data, the resulting environmental footprints can be systematically biased, undermining both academic studies and the policy decisions built upon them.

The consequences of such biases are not trivial. China is the world’s largest manufacturer and exporter, and its carbon neutrality pledge has turned life cycle thinking into a matter of national strategy. Accurate inventories are essential for evaluating whether electric vehicles, photovoltaic panels, batteries, and other low-carbon technologies genuinely deliver net environmental benefits when manufactured in China. They matter for carbon accounting of export goods, for green procurement standards, and for identifying which industrial sectors offer the greatest opportunities for emission reductions. Previous work by members of the team, including a detailed life cycle inventory of Chinese cement production published in Resources, Conservation and Recycling, demonstrated how much insight sector-specific data can provide compared with generic substitutes.

The name of the database is itself a statement of intent. TianGong is named after the seventeenth-century encyclopedia Tian Gong Kai Wu, compiled by Song Yingxing and often translated as The Exploitation of the Works of Nature. That historical text meticulously documented the agricultural and industrial practices of its era, recording how materials were extracted, transformed, and recycled, and it is celebrated today for embodying early concepts of sustainability, ecological balance, and the cyclical use of resources. By invoking this heritage, the developers signal that the database is meant to be a modern counterpart: a systematic, open documentation of how nature’s works are exploited, this time with quantitative rigor and digital accessibility.

The project is also notable for its collaborative structure. The author list spans more than forty institutions, including Tsinghua University, the Chinese Academy of Sciences, Fudan University, Nankai University, Beijing Normal University, Tongji University, and international partners at the University of Groningen, the University of Sheffield, the University of British Columbia, and the University of Adelaide. According to the contribution statements, the work was divided among teams responsible for methodological framework design, sector-specific data collection, data organization, validation, and quality checking, with Ming Xu coordinating the overall research direction and Kebin He providing guidance on the development framework. The effort was supported by the Carbon Neutrality and Energy System Transformation program at Tsinghua University and by grants from the National Natural Science Foundation of China, and it builds on the TianGong Initiative, which supplied rules, guidance, tools, and technical infrastructure.

Access is provided through dedicated platforms, and the developers emphasize community-driven contributions and continuous updates as central to the database’s future. This governance model resembles the approach of established international resources such as ecoinvent, but with a crucial difference: rather than a closed, subscription-based model, TianGong’s MIT licensing invites researchers worldwide to scrutinize, extend, and improve the datasets. Openness of this kind is not merely ideological. Transparent, traceable data allow independent verification of inventory figures, enable meta-analyses across studies, and reduce the duplication of effort that occurs when every research group quietly assembles its own proprietary estimates. For a field that has struggled with reproducibility, a large, documented, openly licensed inventory of the world’s largest industrial economy is a meaningful step forward.

The implications reach well beyond academia. Companies seeking to calculate the carbon footprints of their Chinese supply chains now have a defensible, region-specific foundation for their claims. Regulators designing carbon labeling schemes, green finance taxonomies, or product environmental declarations can ground them in data that reflect actual Chinese production conditions rather than imported assumptions. Researchers modeling decarbonization pathways for steel, cement, chemicals, and power generation can calibrate their models against real process inventories. As the database grows through community contributions and periodic updates, its developers hope it will serve as a reliable foundation for life cycle assessment in China, filling critical data gaps and supporting the resource efficiency and sustainable development goals that motivated the project. In an era when environmental claims face increasing scrutiny, the arrival of a transparent, open, and spatially resolved inventory of Chinese industry may prove to be one of the more consequential infrastructure investments in sustainability science.

Subject of Research: An open-source life cycle inventory database of Chinese industrial unit processes

Article Title: TianGong database—an open-source, ILCD entry-level life cycle unit process database of china

Article References: Xu, M., Chen, W., Wang, Y., Shi, L., Cao, Z., Chang, H., Chen, C., Ding, X., Du, J., Duan, H., Fang, K., Feng, C., Fu, C., Gao, X., Guo, J., Guo, M., He, M., Hu, Y., Huang, B., … He, K. (2026). TianGong database—an open-source, ILCD entry-level life cycle unit process database of china. Journal of Industrial Ecology, 30(4), 2001-2016. https://doi.org/10.1007/s44498-026-00136-7

Image Credits: AI Generated

DOI: 10.1007/s44498-026-00136-7

Keywords: life cycle assessment, TianGong database, China, open-source data, unit process, life cycle inventory, industrial ecology, carbon neutrality, input-output analysis, sustainability, environmental impact assessment, ILCD

The stakes are higher than they might appear. LCA is increasingly deployed not on mature industrial processes but on emerging technologies, from novel photovoltaic materials to experimental hydrogen pathways, where data are scarce, assumptions abound, and modeling choices multiply. When those choices go unreported, other researchers cannot reproduce the work, reviewers cannot properly scrutinize it, and the field accumulates results that cannot be compared or built upon. The scale of the problem is documented in prior reviews: an analysis of a random sample of 319 LCA articles found that only about 40 percent provided a complete and transparent description of their life cycle inventory data, the numerical backbone of any assessment.

The roots of the problem reach back to the founding standards of the method. ISO 14040 describes transparency as an important guiding principle, defining it as the open, comprehensive, and understandable presentation of information, and ISO 14044 specifies items that must be reported on methods, data, and results. But the standards deliberately grant practitioners wide latitude in selecting modeling choices, and that judgment is supposed to be made explicit through reporting and justification. In practice, the new study argues, three obstacles have prevented this: no minimum required information has ever been defined for a study to count as transparent; reporting requirements are scattered across multiple documents, generating duplicates and vague mandates; and the standards have not kept pace with the extra demands of assessing technologies at early stages of maturity.

To build the checklist, the researchers systematically reviewed ISO 14040, 14044, 14025, and 14071, along with the European International Reference Life Cycle Data System Handbook and recent methodological guidance for emerging-technology LCA. They extracted 140 reporting items verbatim, then screened them ruthlessly. Thirty-three were excluded as duplicates, either identical repetitions across documents or semantic duplicates that expressed the same requirement in different words. Twenty-one were dropped as too generic to define a concrete reporting obligation, such as calls for full transparency in value choices without specifying any actual decision to disclose. Twenty-four were judged irrelevant to peer-reviewed research articles, including executive summaries and administrative requirements specific to environmental product declarations. The surviving 62 items were consolidated into 31 checklist entries, each carrying reporting guidance, a location field for authors to indicate where the information appears, and a pass-or-fail field for reviewers.

Twenty-four of the 31 items are obligatory for every LCA study, covering fundamentals such as the functional unit, reference flow, system boundary, foreground and background system definitions, data sources, impact assessment method, and limitations. The remaining seven are conditional, triggered only by specific study contexts. An allocation procedure must be reported only when a product system is multifunctional; a scale-up method only when lab data are extrapolated to industrial scale; calculation steps only when inventory flows are computed or transformed rather than taken directly from databases. The decision rule is deliberately non-compensatory: a study passes only if it reports all obligatory items and all applicable conditional ones. The logic, the authors explain, follows directly from the ISO definition, since a study missing any required information is neither fully open nor comprehensively reported, and therefore cannot be called transparent.

To test whether the checklist could actually be applied, the team screened 370 records from a Scopus search and assessed ten recently published LCA case studies of emerging technologies. The results were revealing. Core items such as the reasons for the study, system boundary, data sources, and limitations were reported in all ten articles, and sensitivity analysis in nine. But two items drawn directly from ISO 14044 received zero passes across the entire sample: the description of previous iterations of the goal and scope, and cut-off criteria, the rules governing which flows and processes were excluded from the model. Their systematic omission, the authors note, may have implications for those studies’ implicit claims of ISO alignment. Other items fared barely better: temporal scope was reported in only three of ten studies, targeted audience in four, and completeness and consistency checks in one and two respectively.

The team then demonstrated what full compliance looks like by conducting a complete LCA of an emerging photovoltaic material, a silver bismuth sulfide film prepared by one-step solution crystallization, and reporting it against every checklist item. The case study is technically instructive in its own right. The functional unit was defined as one kilowatt-hour of electricity generated in southern Europe at 5.15 percent power conversion efficiency, requiring a reference flow of about 24 milligrams of film. The cradle-to-gate assessment found that producing that film at laboratory scale, where spin coating wastes over 99 percent of the precursor solution, emits 0.28 kilograms of carbon dioxide equivalent, close to the footprint of simply buying a kilowatt-hour of European grid electricity.

Contribution analysis pinpointed the hotspots with striking clarity: silver nitrate dominates most impact categories, with more than 99 percent of that impact traceable to silver itself, while bismuth nitrate, driven by bismuth metal production, ranks second. Untreated process emissions, including nitrates, nitrogen oxides, and thiourea, dominate three toxicity-related categories. When the researchers modeled industrial-scale deposition using slot-die coating with material utilization rates of 80 or 90 percent, instead of the current 0.69 percent achieved in the lab, climate impacts fell roughly 98-fold, to around 0.005 kilograms of carbon dioxide equivalent. The conclusion for solar researchers is blunt: the absorber material’s proof of concept is secure, but future work should prioritize replacing silver or designing cells that allow its recovery.

The authors are candid about the limits of their proposal. The checklist defines a minimum reporting requirement, not a complete measure of scientific transparency or methodological rigor; it cannot predefine study-specific limitations, and some items remain generic across study contexts. Different scale-up methods, for instance, demand different forms of documentation, from fully reported engineering equations to access to simulation models, and the checklist does not yet specify which. The authors suggest that large language models could eventually help by pre-filling the checklist from a manuscript, leaving authors to verify the output, and they recommend that journals adapt the tool for peer review, drawing on the structured verification logic already used by environmental product declaration programs.

What the study ultimately offers is a paradigm shift in how the field talks about openness. Transparency has long functioned as an aspiration, invoked in papers and standards but never tested. By converting it into a pass-or-fail threshold with traceable lineage to ISO requirements, the Bordeaux-led team gives authors a concrete target, reviewers a consistent yardstick, and editors a potential instrument for screening submissions. Adoption will require coordinated effort from standards bodies, researchers, and journal editors, and the checklist itself is presented as a starting point rather than a final word. But for a method whose credibility rests on whether its assumptions and data choices can be seen, the message lands: a study that hides its modeling decisions is not slightly less transparent. It is not transparent at all.

Subject of Research: A minimum transparency reporting threshold and checklist for life cycle assessment of emerging technologies

Article Title: Toward a Minimum Threshold for Transparent Reporting in LCA of Emerging Technologies: A Checklist

Article References: Kamali, A. K., Cojocaru, L., Toupance, T., Kubo, T., Segawa, H., Laratte, B., & Sonnemann, G. (2026). Toward a Minimum Threshold for Transparent Reporting in LCA of Emerging Technologies: A Checklist. Cleaner Engineering and Technology, 34, Article 101323. https://doi.org/10.1016/j.clet.2026.101323

Image Credits: AI Generated

DOI: 10.1016/j.clet.2026.101323

Keywords: life cycle assessment, transparency, ISO 14040, emerging technologies, reporting checklist, peer review, photovoltaics, silver bismuth sulfide, life cycle inventory, sustainability, functional unit, scale-up